Adding Gemini CLI

* anthropic provider

Co-authored-by: StvLz <lizarazo.steven@gmail.com>

* base

---------

Co-authored-by: StvLz <lizarazo.steven@gmail.com>

.changeset/chatty-turkeys-wait.md

@@ -1,5 +0,0 @@
----
-"claude-dev": patch
----
-
-fix: respect setting litellm models for plan and act

.changeset/curvy-bulldogs-jog.md

@@ -1,5 +0,0 @@
----
-"claude-dev": minor
----
-
-Migrate didBecomeVisible to protobus

.changeset/eight-jokes-punch.md

@@ -1,5 +0,0 @@
----
-"claude-dev": patch
----
-
-Do not clear the input value when changing modes without sending a message

.changeset/lazy-cherries-search.md

@@ -1,5 +0,0 @@
----
-"claude-dev": patch
----
-
-Prevent reading IS_DEV from the users environment

.changeset/pretty-bikes-tickle.md

@@ -1,5 +0,0 @@
----
-"claude-dev": patch
----
-
-fix(bedrock): remove custom Model encode

.changeset/seven-zebras-grin.md

@@ -1,5 +0,0 @@
----
-"claude-dev": patch
----
-
-Updated Timeline tooltips for `followup` messages and consolidated color retrieval code.

.changeset/six-walls-doubt.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+telemtrySetting protobus migration

.changeset/spicy-carpets-begin.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+clearAllTaskHistory protobus migration

.changeset/thirty-donkeys-whisper.md

@@ -1,5 +0,0 @@
----
-"claude-dev": patch
----
-
-The close task button and delete task button in the task header are now correctly announced by screen readers.

.clinerules/protobuf-development.md

@@ -0,0 +1,89 @@
+# Cline Protobuf Development Guide
+
+This guide outlines how to add new gRPC endpoints for communication between the webview (frontend) and the extension host (backend).
+
+## Overview
+
+Cline uses [Protobuf](https://protobuf.dev/) to define a strongly-typed API, ensuring efficient and type-safe communication. All definitions are in the `/proto` directory. The compiler and plugins are included as project dependencies, so no manual installation is needed.
+
+## Key Concepts & Best Practices
+
+-   **File Structure**: Each feature domain should have its own `.proto` file (e.g., `account.proto`, `task.proto`).
+-   **Message Design**:
+    -   For simple, single-value data, use the shared types in `proto/common.proto` (e.g., `StringRequest`, `Empty`, `Int64Request`). This promotes consistency.
+    -   For complex data structures, define custom messages within the feature's `.proto` file (see `task.proto` for examples like `NewTaskRequest`).
+-   **Naming Conventions**:
+    -   Services: `PascalCaseService` (e.g., `AccountService`).
+    -   RPCs: `camelCase` (e.g., `accountEmailIdentified`).
+    -   Messages: `PascalCase` (e.g., `StringRequest`).
+-   **Streaming**: For server-to-client streaming, use the `stream` keyword on the response type. See `subscribeToAuthCallback` in `account.proto` for an example.
+
+---
+
+## 4-Step Development Workflow
+
+Here’s how to add a new RPC, using `scrollToSettings` as an example.
+
+### 1. Define the RPC in a `.proto` File
+
+Add your service method to the appropriate file in the `proto/` directory.
+
+**File: `proto/ui.proto`**
+```proto
+service UiService {
+  // ... other RPCs
+  // Scrolls to a specific settings section in the settings view
+  rpc scrollToSettings(StringRequest) returns (KeyValuePair);
+}
+```
+Here, we use the common `StringRequest` and `KeyValuePair` types.
+
+### 2. Compile Definitions
+
+After editing a `.proto` file, regenerate the TypeScript code. From the project root, run:
+```bash
+npm run protos
+```
+This command compiles all `.proto` files and outputs the generated code to `src/generated/` and `src/shared/`. Do not edit these generated files manually.
+
+### 3. Implement the Backend Handler
+
+Create the RPC implementation in the backend. Handlers are located in `src/core/controller/[service-name]/`.
+
+**File: `src/core/controller/ui/scrollToSettings.ts`**
+```typescript
+import { Controller } from ".."
+import { StringRequest, KeyValuePair } from "../../../shared/proto/common"
+
+/**
+ * Executes a scroll to settings action
+ * @param controller The controller instance
+ * @param request The request containing the ID of the settings section to scroll to
+ * @returns KeyValuePair with action and value fields for the UI to process
+ */
+export async function scrollToSettings(controller: Controller, request: StringRequest): Promise<KeyValuePair> {
+	return KeyValuePair.create({
+		key: "scrollToSettings",
+		value: request.value || "",
+	})
+}
+```
+
+### 4. Call the RPC from the Webview
+
+Call the new RPC from a React component in `webview-ui/`. The generated client makes this simple.
+
+**File: `webview-ui/src/components/browser/BrowserSettingsMenu.tsx`** (Example)
+```tsx
+import { UiServiceClient } from "../../../services/grpc"
+import { StringRequest } from "../../../../shared/proto/common"
+
+// ... inside a React component
+const handleMenuClick = async () => {
+    try {
+        await UiServiceClient.scrollToSettings(StringRequest.create({ value: "browser" }))
+    } catch (error) {
+        console.error("Error scrolling to browser settings:", error)
+    }
+}
+```

.github/ISSUE_TEMPLATE/feature_contribution.yml

@@ -0,0 +1,116 @@
+name: 💡 Feature Proposal & Contribution
+description: Propose a new feature or improvement, and optionally offer to implement feature as a contributor
+labels: ["proposal"]
+body:
+    - type: markdown
+      attributes:
+          value: |
+              **Feature Proposal & Contribution for Cline**
+
+              Thank you for proposing a feature or improvement for Cline! This template helps us understand the problem, evaluate the solution, and coordinate implementation.
+
+              **For detailed proposals:** Please provide comprehensive information to enable fast prioritization and discussion.
+              **For contribution offers:** You can indicate your willingness to implement the feature yourself.
+
+              Before submitting:
+              - Search existing [Issues](https://github.com/cline/cline/issues) and [Discussions](https://github.com/cline/cline/discussions) to avoid duplicates
+              - Read the [Contributing Guide](https://github.com/cline/cline/blob/main/CONTRIBUTING.md) if you plan to contribute
+              - Don't start implementation until the proposal is reviewed and approved
+
+    - type: textarea
+      id: problem-description
+      attributes:
+          label: What problem does this solve?
+          description: |
+              Describe the problem clearly from a user's point of view. Focus on why this matters, who it affects, and when it occurs.
+
+              ✅ Good examples:
+              - "LLM provider returns 400 error when nearing the context window instead of truncating"
+              - "Submit button is invisible in dark mode"
+              - "Users can't easily share their Cline configurations with team members"
+
+              ❌ Avoid vague descriptions:
+              - "Performance is bad"
+              - "UI needs work"
+
+              Your description should include:
+              - Who is affected?
+              - When does it happen?
+              - What's the current vs expected behavior?
+              - What is the impact?
+          placeholder: Be specific about the problem, who it affects, and the impact.
+      validations:
+          required: true
+
+    - type: textarea
+      id: proposed-solution
+      attributes:
+          label: What's the proposed solution?
+          description: |
+              Describe how the problem should be solved. Be specific about UX, system behavior, and any flows that would change.
+
+              ✅ Good examples:
+              - "Add error handling immediately after attempting to create the llm stream and retry after manually truncating"
+              - "Update button styling to ensure contrast in all themes"
+              - "Add export/import functionality in settings with JSON format"
+
+              ❌ Avoid vague solutions:
+              - "Improve performance"
+              - "Fix the bug"
+
+              Your solution should include:
+              - What exactly will change?
+              - How will users interact with it?
+              - What's the expected outcome?
+          placeholder: Describe the proposed changes and how they solve the problem.
+      validations:
+          required: false
+
+    - type: dropdown
+      id: contribution-intent
+      attributes:
+          label: Are you interested in implementing this?
+          description: Let us know if you'd like to contribute to this feature
+          options:
+              - "No, just proposing the idea"
+              - "Yes, I'd like to implement this myself"
+              - "Yes, I'd like to collaborate with others"
+              - "Maybe, depending on complexity and guidance"
+      validations:
+          required: false
+
+    - type: textarea
+      id: implementation-approach
+      attributes:
+          label: Implementation approach (if contributing)
+          description: |
+              **Only fill this out if you selected "Yes" above.**
+
+              How do you plan to implement this? Include:
+              - High-level technical approach
+              - Files/components that would be affected
+              - Any new dependencies required
+              - Potential challenges or considerations you've identified
+
+              This helps us provide better guidance and ensures alignment before you start coding.
+          placeholder: "My implementation approach would be..."
+
+    - type: checkboxes
+      id: checklist
+      attributes:
+          label: Proposal checklist
+          options:
+              - label: I've checked for existing issues or related proposals
+                required: true
+              - label: I understand this needs review before implementation can start
+                required: true
+
+    - type: checkboxes
+      id: contribution-checklist
+      attributes:
+          label: Contribution checklist (if contributing)
+          description: Only check these if you plan to contribute
+          options:
+              - label: I've read the [Contributing Guide](https://github.com/cline/cline/blob/main/CONTRIBUTING.md)
+              - label: I'm willing to make changes based on feedback
+              - label: I understand the code review process and requirements

.github/ISSUE_TEMPLATE/feature_proposal.yml

@@ -1,130 +0,0 @@
-name: 📝 Detailed Feature Proposal
-description: Propose a new feature or improvement
-labels: ["proposal"]
-body:
-    - type: markdown
-      attributes:
-          value: |
-              **Feature Proposal for Cline**
-
-              Thank you for creating a feature proposal for Cline! This template is for clear, actionable proposals that define a specific problem and a high-confidence solution. Please provide enough detail to enable fast prioritization, discussion, and execution.
-
-              Detailed proposals will be prioritized, while vague proposals may be closed or require extensive back and forth communication.
-
-              Before submitting:
-              - Search existing [Issues](https://github.com/cline/cline/issues) and [Discussions](https://github.com/cline/cline/discussions) to avoid duplicates
-              - Don’t start implementation until the proposal is reviewed and approved
-
-    - type: textarea
-      id: problem-description
-      attributes:
-          label: What problem does this solve?
-          description: |
-              Describe the problem clearly from a user's point of view. Focus on why this matters, who it affects, and when it occurs.
-
-              ✅ Solid:
-              - "LLM provider returns 400 error when nearing the context window instead of truncating"
-              - "Submit button is invisible in dark mode"
-
-              ❌ Avoid:
-              - "Performance is bad"
-              - "UI needs work"
-
-              Your description should include:
-              - Who is affected?
-              - When does it happen?
-              - What's the current vs expected behavior?
-              - What is the impact?
-          placeholder: Be specific about the problem, who it affects, and the impact.
-      validations:
-          required: true
-
-    - type: textarea
-      id: proposed-solution
-      attributes:
-          label: What’s the proposed solution?
-          description: |
-              Describe how the problem should be solved. Be specific about UX, system behavior, and any flows that would change.
-
-              ✅ Solid:
-              - "Add error handling immediately after attempting to create the llm stream and retry after manually truncating"
-              - "Update button styling to ensure contrast in all themes"
-
-              ❌ Avoid:
-              - "Improve performance"
-              - "Fix the bug"
-
-              Your solution should include:
-              - What exactly will change?
-              - How will users interact with it?
-              - What’s the expected outcome?
-          placeholder: Describe the proposed changes and how they solve the problem.
-      validations:
-          required: true
-
-    - type: textarea
-      id: acceptance-criteria
-      attributes:
-          label: How will we know it works? (Acceptance Criteria)
-          description: Define clear, testable success criteria.
-          placeholder: Provide specific and testable conditions for success.
-      validations:
-          required: true
-
-    - type: textarea
-      id: estimated-effort
-      attributes:
-          label: Estimated effort and complexity
-          description: |
-              Help us understand scope and risks. Include:
-
-              - Size estimate (XS/S/M/L/XL or hours/days)
-              - Why this size? What’s technically involved?
-              - Any tricky parts, refactors, or risks?
-              - Performance or compatibility concerns?
-              - Any dependencies on systems, teams, or libraries?
-          placeholder: Size, reasoning, risks, and dependencies.
-      validations:
-          required: true
-
-    - type: textarea
-      id: technical-considerations
-      attributes:
-          label: Technical considerations, tradeoffs, and/or risks (optional)
-          description: |
-              Include any technical context that helps us evaluate or implement the proposal more effectively.
-
-              You may include:
-              - Architectural changes or required refactors
-              - Performance implications or system-level impacts
-              - Known limitations, risks, or tricky edge cases
-              - Compatibility concerns or migration steps
-              - Alternative approaches you considered and why they were not chosen
-              - Dependencies on other systems, teams, or libraries
-              - Were other approaches considered? Why is this one preferred?
-          placeholder: Technical considerations, tradeoffs, and/or risks.
-
-    - type: textarea
-      id: additional-context
-      attributes:
-          label: Additional context (optional)
-          description: Diagrams, mockups, logs, links, or anything else that helps explain or justify the proposal.
-          placeholder: Diagrams, mockups, logs, links, or anything else that helps explain or justify the proposal.
-
-    - type: checkboxes
-      id: checklist
-      attributes:
-          label: Proposal checklist
-          options:
-              - label: I’ve checked for existing issues or related proposals
-                required: true
-              - label: I understand this needs review before implementation can start
-                required: true
-
-    - type: checkboxes
-      id: willingness-to-contribute
-      attributes:
-          label: Interested in implementing this?
-          description: Optional
-          options:
-              - label: Yes, I’d like to help implement this feature

.github/workflows/stale.yml

@@ -12,7 +12,7 @@ jobs:
             issues: write
             pull-requests: write
         steps:
-            - uses: actions/stale@28ca103
+            - uses: actions/stale@v9
               with:
                   days-before-issue-stale: 60
                   days-before-issue-close: 14

.gitignore

@@ -35,4 +35,7 @@ webview-ui/src/services/grpc-client.ts
 # Host bridge
 src/hosts/vscode/*/methods.ts
 src/hosts/vscode/*/index.ts
+src/hosts/vscode/client/host-grpc-client.ts
 src/hosts/vscode/host-grpc-service-config.ts
+src/standalone/server-setup.ts
+gemini-cli

.prettierignore

@@ -5,3 +5,4 @@ webview-ui/build/
 package-lock.json
 src/core/prompts/system.ts
 src/core/prompts/model_prompts/claude4.ts
+evals/

.vscode/launch.json

@@ -50,10 +50,11 @@
 			"outFiles": ["${workspaceFolder}/dist-standalone/**/*.js"],
 			"preLaunchTask": "compile-standalone",
 			"env": {
-				"GRPC_TRACE": "all",
-				"GRPC_VERBOSITY": "DEBUG",
+				// Turns on grpc debug log.
+				//"GRPC_TRACE": "all",
+				//"GRPC_VERBOSITY": "DEBUG",
 				"NODE_PATH": "${workspaceFolder}/dist-standalone/node_modules",
-				"CLINE_DIR": "${userHome}/.cline-standalone",
+
 				"HOST_BRIDGE_ADDRESS": "localhost:50052"
 			},
 			"program": "standalone.js"

CHANGELOG.md

@@ -1,5 +1,21 @@
 # Changelog
 
+## [3.17.16]
+
+-   Fix Claude Code provider error handling for incomplete messages during long-running tasks (Thanks @BarreiroT!)
+-   Add taskId as metadata to LiteLLM API requests for better request tracing (Thanks @jorgegarciarey!)
+
+## [3.17.15]
+
+-   Fix LiteLLM provider to properly respect selected model IDs when switching between Plan and Act modes (Thanks @sammcj!)
+-   Fix chat input being cleared when switching between Plan/Act modes without sending a message (Thanks @BarreiroT!)
+-   Fix MCP server name display to avoid showing "undefined" for SSE servers, preventing tool/resource invocation failures (Thanks @ramybenaroya!)
+-   Fix AWS Bedrock provider by removing deprecated custom model encoding (Thanks @watany-dev!)
+-   Fix timeline tooltips for followup messages and improve color retrieval code (Thanks @char8x!)
+-   Improve accessibility by making task header buttons properly announced by screen readers (Thanks @yncat!)
+-   Improve accessibility by adding proper state reporting for Plan/Act mode switch for screen readers (Thanks @yncat!)
+-   Prevent reading development environment variables from user's environment (Thanks @BarreiroT!)
+
 ## [3.17.14]
 
 -   Add Claude Code as a new API provider, allowing integration with Anthropic's Claude Code CLI tool and Claude Max Plan (Thanks @BarreiroT!)

CONTRIBUTING.md

@@ -10,13 +10,6 @@ Bug reports help make Cline better for everyone! Before creating a new issue, pl
      🔐 <b>Important:</b> If you discover a security vulnerability, please use the <a href="https://github.com/cline/cline/security/advisories/new">Github security tool to report it privately</a>.
 </blockquote>
 
-## Deciding What to Work On
-
-Looking for a good first contribution? Check out issues labeled ["good first issue"](https://github.com/cline/cline/labels/good%20first%20issue) or ["help wanted"](https://github.com/cline/cline/labels/help%20wanted). These are specifically curated for new contributors and areas where we'd love some help!
-
-We also welcome contributions to our [documentation](https://github.com/cline/cline/tree/main/docs)! Whether it's fixing typos, improving existing guides, or creating new educational content - we'd love to build a community-driven repository of resources that helps everyone get the most out of Cline. You can start by diving into `/docs` and looking for areas that need improvement.
-
-If you're planning to work on a bigger feature, please create a [feature request](https://github.com/cline/cline/discussions/categories/feature-requests?discussions_q=is%3Aopen+category%3A%22Feature+Requests%22+sort%3Atop) first so we can discuss whether it aligns with Cline's vision.
 
 ## Before Contributing
 
@@ -24,14 +17,70 @@ All contributions must begin with a GitHub Issue, unless the change is for small
 
 - **Check existing issues**: Search [GitHub Issues](https://github.com/cline/cline/issues).
 - **Create an issue**: Use appropriate templates:
-    - **Bugs:** "Bug Report" template.
-    - **Features:** "Detailed Feature Proposal" template. Approval from a core Cline contributor required before starting.
-- **Claim issues**: Comment your interest.
+    - **Contributions:** Use the "Contribution Request" template to propose what you'd like to work on.
+    - **Bugs:** "Bug Report" template for reporting issues.
+    - **Features:** "Detailed Feature Proposal" template for suggesting new features.
+- **Wait for approval**: A core Cline contributor must approve your contribution request before you start implementation.
+- **Claim issues**: Once approved, the issue will be assigned to you.
 
 **PRs without approved issues may be closed.**
 
+
+## Deciding What to Work On
+
+Looking for a good first contribution? Check out issues labeled ["good first issue"](https://github.com/cline/cline/labels/good%20first%20issue) or ["help wanted"](https://github.com/cline/cline/labels/help%20wanted). These are specifically curated for new contributors and areas where we'd love some help!
+
+We also welcome contributions to our [documentation](https://github.com/cline/cline/tree/main/docs)! Whether it's fixing typos, improving existing guides, or creating new educational content - we'd love to build a community-driven repository of resources that helps everyone get the most out of Cline. You can start by diving into `/docs` and looking for areas that need improvement.
+
 ## Development Setup
 
+
+### Local Development Instructions
+
+1. Clone the repository _(Requires [git-lfs](https://git-lfs.com/))_:
+    ```bash
+    git clone https://github.com/cline/cline.git
+    ```
+2. Open the project in VSCode:
+    ```bash
+    code cline
+    ```
+3. Install the necessary dependencies for the extension and webview-gui:
+    ```bash
+    npm run install:all
+    ```
+4. Launch by pressing `F5` (or `Run`->`Start Debugging`) to open a new VSCode window with the extension loaded. (You may need to install the [esbuild problem matchers extension](https://marketplace.visualstudio.com/items?itemName=connor4312.esbuild-problem-matchers) if you run into issues building the project.)
+
+
+
+
+### Creating a Pull Request
+
+1. Before creating a PR, generate a changeset entry:
+    ```bash
+    npm run changeset
+    ```
+   This will prompt you for:
+   - Type of change (major, minor, patch)
+     - `major` → breaking changes (1.0.0 → 2.0.0)
+     - `minor` → new features (1.0.0 → 1.1.0)
+     - `patch` → bug fixes (1.0.0 → 1.0.1)
+   - Description of your changes
+
+2. Commit your changes and the generated `.changeset` file
+
+3. Push your branch and create a PR on GitHub. Our CI will:
+   - Run tests and checks
+   - Changesetbot will create a comment showing the version impact
+   - When merged to main, changesetbot will create a Version Packages PR
+   - When the Version Packages PR is merged, a new release will be published
+4. Testing
+    - Run `npm run test` to run tests locally. 
+    - Before submitting PR, run `npm run format:fix` to format your code
+    - Run `npm run test:ci` to run tests locally
+
+### Extension
+
 1. **VS Code Extensions**
 
     - When opening the project, VS Code will prompt you to install recommended extensions
@@ -41,6 +90,7 @@ All contributions must begin with a GitHub Issue, unless the change is for small
 2. **Local Development**
     - Run `npm run install:all` to install dependencies
     - Run `npm run test` to run tests locally
+    - Run → Start Debugging or `>Debug: Select and Start Debugging` and wait for a new VS Code instance to open
     - Before submitting PR, run `npm run format:fix` to format your code
 
 3. **Linux-specific Setup**
@@ -85,8 +135,6 @@ All contributions must begin with a GitHub Issue, unless the change is for small
       xvfb
     ```
 
-    - Run `npm run test:ci` to run tests locally
-
 ## Writing and Submitting Code
 
 Anyone can contribute code to Cline, but we ask that you follow these guidelines to ensure your contributions can be smoothly integrated:

README.md

@@ -141,50 +141,6 @@ For example, when working with a local web server, you can use 'Restore Workspac
 
 To contribute to the project, start with our [Contributing Guide](CONTRIBUTING.md) to learn the basics. You can also join our [Discord](https://discord.gg/cline) to chat with other contributors in the `#contributors` channel. If you're looking for full-time work, check out our open positions on our [careers page](https://cline.bot/join-us)!
 
-<details>
-<summary>Local Development Instructions</summary>
-
-1. Clone the repository _(Requires [git-lfs](https://git-lfs.com/))_:
-    ```bash
-    git clone https://github.com/cline/cline.git
-    ```
-2. Open the project in VSCode:
-    ```bash
-    code cline
-    ```
-3. Install the necessary dependencies for the extension and webview-gui:
-    ```bash
-    npm run install:all
-    ```
-4. Launch by pressing `F5` (or `Run`->`Start Debugging`) to open a new VSCode window with the extension loaded. (You may need to install the [esbuild problem matchers extension](https://marketplace.visualstudio.com/items?itemName=connor4312.esbuild-problem-matchers) if you run into issues building the project.)
-
-</details>
-
-<details>
-<summary>Creating a Pull Request</summary>
-
-1. Before creating a PR, generate a changeset entry:
-    ```bash
-    npm run changeset
-    ```
-   This will prompt you for:
-   - Type of change (major, minor, patch)
-     - `major` → breaking changes (1.0.0 → 2.0.0)
-     - `minor` → new features (1.0.0 → 1.1.0)
-     - `patch` → bug fixes (1.0.0 → 1.0.1)
-   - Description of your changes
-
-2. Commit your changes and the generated `.changeset` file
-
-3. Push your branch and create a PR on GitHub. Our CI will:
-   - Run tests and checks
-   - Changesetbot will create a comment showing the version impact
-   - When merged to main, changesetbot will create a Version Packages PR
-   - When the Version Packages PR is merged, a new release will be published
-
-</details>
-
-
 ## License
 
 [Apache 2.0 © 2025 Cline Bot Inc.](./LICENSE)

docs/docs.json

@@ -146,6 +146,7 @@
 				"group": "Provider Configuration",
 				"pages": [
 					"provider-config/anthropic",
+					"provider-config/claude-code",
 					"provider-config/aws-bedrock-with-credentials-authentication",
 					"provider-config/aws-bedrock-with-profile-authentication",
 					"provider-config/gcp-vertex-ai",

docs/gemini-cli-provider.md

@@ -0,0 +1,93 @@
+# Gemini CLI Provider for Cline
+
+This document describes how to use the Gemini CLI as a provider in Cline.
+
+## Overview
+
+The Gemini CLI provider allows you to use Google's Gemini models through the Gemini CLI tool instead of using API keys directly. This can be useful if you:
+
+- Already have the Gemini CLI installed and configured
+- Want to use the CLI's authentication methods (OAuth, etc.)
+- Prefer to manage your Gemini access through the CLI
+
+## Prerequisites
+
+1. **Install the Gemini CLI**: The Gemini CLI must be installed and accessible from your system PATH.
+   
+2. **Set up authentication**: The CLI needs to be authenticated. You can either:
+   - Set the `GEMINI_API_KEY` environment variable
+   - Use the CLI's built-in authentication methods
+
+3. **Build the CLI** (if using from source):
+   ```bash
+   cd gemini-cli/packages/cli
+   npm install
+   npm run build
+   ```
+
+## Configuration
+
+To use the Gemini CLI provider in Cline:
+
+1. Open Cline settings
+2. Select "Gemini CLI" as your API provider
+3. (Optional) Specify the path to the Gemini CLI executable if it's not in your PATH
+4. Select your desired Gemini model
+
+## Supported Models
+
+The Gemini CLI provider supports the same models as the regular Gemini provider:
+
+- gemini-2.5-pro
+- gemini-2.5-flash
+- gemini-2.0-flash-001
+- gemini-1.5-flash-002
+- gemini-1.5-pro-002
+
+## How It Works
+
+1. When you send a message in Cline, it converts your conversation to a format the Gemini CLI understands
+2. Cline spawns the Gemini CLI as a subprocess with your prompt
+3. The CLI's response is streamed back to Cline
+4. The response is displayed in the Cline interface
+
+## Limitations
+
+- **No image support**: The Gemini CLI in non-interactive mode doesn't support images
+- **No tool execution**: The CLI handles its own tool execution, which doesn't integrate with Cline's tools
+- **Token counting**: Token usage is estimated based on text length rather than actual token counts
+- **No caching**: The CLI doesn't support Anthropic-style prompt caching
+
+## Troubleshooting
+
+### CLI not found
+- Ensure the Gemini CLI is installed and in your PATH
+- Or specify the full path to the CLI in Cline settings
+
+### Authentication errors
+- Check that `GEMINI_API_KEY` is set in your environment
+- Or ensure the CLI is authenticated using its built-in methods
+
+### No response
+- Try running the CLI manually to ensure it works:
+  ```bash
+  gemini --prompt "Hello, world!"
+  ```
+
+### Build errors
+If using from source, ensure the CLI is built:
+```bash
+cd gemini-cli/packages/cli
+npm install
+npm run build
+```
+
+## Testing
+
+You can test the integration using the provided test script:
+
+```bash
+node test-gemini-cli.js
+```
+
+This will attempt to run a simple prompt through the CLI and report whether it succeeded.

docs/provider-config/claude-code.mdx

@@ -0,0 +1,86 @@
+---
+title: "Claude Code"
+description: "Use your Claude Max subscription with Cline instead of paying per token. Learn how to set up and configure the Claude Code provider."
+---
+
+**Website:** [https://docs.anthropic.com/en/docs/claude-code/setup](https://docs.anthropic.com/en/docs/claude-code/setup)
+
+The Claude Code provider lets you use your existing Claude subscription with Cline. If you have Claude Max, this means you can use Claude in Cline without paying extra API costs.
+
+<Frame>
+	<img
+		src="https://storage.googleapis.com/cline_public_images/docs/assets/claude-code-use-opus.gif"
+		alt="Using the Claude Code provider in Cline with Opus model"
+	/>
+</Frame>
+
+## Setup
+
+First, you'll need to install and authenticate Claude Code on your system:
+
+1. **Install Claude Code**: Follow Anthropic's [official setup guide](https://docs.anthropic.com/en/docs/claude-code/setup) to install and authenticate the Claude CLI.
+
+2. **Configure in Cline**:
+    - Open Cline settings (⚙️ icon)
+    - Select **Claude Code** from the **API Provider** dropdown
+    - Set the path to your Claude CLI executable (usually just `claude` if it's in your PATH)
+
+<Frame>
+	<img
+		src="https://storage.googleapis.com/cline_public_images/docs/assets/claude-code-setup.gif"
+		alt="Setting up the Claude Code provider in Cline"
+	/>
+</Frame>
+
+### Finding your Claude Code path
+
+If you're not sure where Claude Code is installed:
+
+-   **macOS / Linux**: Run `which claude` in your terminal
+-   **Windows (Command Prompt)**: Run `where claude`
+-   **Windows (PowerShell)**: Run `Get-Command claude`
+
+## Supported Models
+
+The Claude Code provider supports these models:
+
+-   `claude-sonnet-4-20250514` (Recommended)
+-   `claude-opus-4-20250514`
+-   `claude-3-7-sonnet-20250219`
+-   `claude-3-5-sonnet-20241022`
+-   `claude-3-5-haiku-20241022`
+
+## How it works
+
+When you use Claude Code with Cline, here's what happens behind the scenes:
+
+Cline wraps the Claude Code CLI to handle your requests. Each time you send a message, Cline starts a new `claude` process, sends your conversation, and streams the response back. The AI reasoning comes from Claude Code, but all the actual file editing, terminal commands, and other tools are handled by Cline.
+
+The main difference you'll notice is that responses don't stream character-by-character like other providers. Instead, Claude Code processes your full request before sending back the complete response.
+
+## Limitations
+
+There are a few things to keep in mind with Claude Code:
+
+-   Images in your messages get converted to text placeholders since Claude Code doesn't support image uploads through the CLI
+-   Prompt caching isn't available with this provider
+-   Responses don't stream in real-time like other providers
+
+## Troubleshooting
+
+If you run into issues:
+
+**Authentication problems**: Make sure you're logged into Claude Code with your subscription account. Run `claude auth status` to check.
+
+**Path issues**: Double-check that the Claude CLI path in Cline's settings is correct. Try running `claude --version` in your terminal to verify it's working.
+
+**Still having trouble?** We're actively improving this integration. Report issues on our [GitHub](https://github.com/cline/cline/issues) or ask for help in our [Discord](https://discord.gg/cline).
+
+## Usage with subscriptions
+
+If you have a Claude Max subscription, your usage in Cline shows up as $0.00 in the billing interface since you're not paying additional API costs. Your usage still counts against your subscription limits, but you won't see per-token charges.
+
+For more details about using Claude Code with your subscription, check out Anthropic's documentation:
+
+-   [Claude Code Setup Guide](https://docs.anthropic.com/en/docs/claude-code/setup)
+-   [Using Claude Code with Pro/Max Plans](https://support.anthropic.com/en/articles/11145838-using-claude-code-with-your-pro-or-max-plan)

eslint-rules/__tests__/no-vscode-postmessage.test.ts

@@ -0,0 +1,74 @@
+const { RuleTester: VscodeRuleTester } = require("eslint")
+const vscodePostmessageRule = require("../no-vscode-postmessage")
+
+const vscodeRuleTester = new VscodeRuleTester({
+	parser: require.resolve("@typescript-eslint/parser"),
+	parserOptions: {
+		ecmaVersion: 2020,
+		sourceType: "module",
+		ecmaFeatures: {
+			jsx: true,
+		},
+	},
+})
+
+vscodeRuleTester.run("no-vscode-postmessage", vscodePostmessageRule, {
+	valid: [
+		// Should allow vscode.postMessage in grpc-client-base.ts
+		{
+			code: `vscode.postMessage({ type: "grpc_request", data: {} })`,
+			filename: "grpc-client-base.ts",
+		},
+		{
+			code: `vscode.postMessage({ type: "grpc_request_cancel" })`,
+			filename: "/path/to/grpc-client-base.ts",
+		},
+		// Should allow other vscode API calls
+		{
+			code: `vscode.window.showInformationMessage("Hello")`,
+			filename: "test.ts",
+		},
+		// Should allow postMessage calls on other objects
+		{
+			code: `window.postMessage({ type: "test" }, "*")`,
+			filename: "test.ts",
+		},
+		// Should allow variables named vscode but not calling postMessage
+		{
+			code: `const vscode = { other: "method" }; vscode.other()`,
+			filename: "test.ts",
+		},
+	],
+	invalid: [
+		// Should ban vscode.postMessage in regular files
+		{
+			code: `vscode.postMessage({ type: "test", data: {} })`,
+			filename: "test.ts",
+			errors: [
+				{
+					messageId: "useGrpcClient",
+				},
+			],
+		},
+		// Should ban vscode.postMessage in components
+		{
+			code: `vscode.postMessage({ type: "apiConfiguration", apiConfiguration })`,
+			filename: "ApiOptions.tsx",
+			errors: [
+				{
+					messageId: "useGrpcClient",
+				},
+			],
+		},
+		// Should ban vscode.postMessage in test files
+		{
+			code: `vscode.postMessage({ type: "newTask", text: message.text })`,
+			filename: "test.test.ts",
+			errors: [
+				{
+					messageId: "useGrpcClient",
+				},
+			],
+		},
+	],
+})

eslint-rules/index.js

@@ -1,11 +1,13 @@
 // eslint-rules/index.js
 const noProtobufObjectLiterals = require("./no-protobuf-object-literals")
 const noGrpcClientObjectLiterals = require("./no-grpc-client-object-literals")
+const noVscodePostmessage = require("./no-vscode-postmessage")
 
 module.exports = {
 	rules: {
 		"no-protobuf-object-literals": noProtobufObjectLiterals,
 		"no-grpc-client-object-literals": noGrpcClientObjectLiterals,
+		"no-vscode-postmessage": noVscodePostmessage,
 	},
 	configs: {
 		recommended: {
@@ -13,6 +15,7 @@ module.exports = {
 			rules: {
 				"local/no-protobuf-object-literals": "error",
 				"local/no-grpc-client-object-literals": "error",
+				"local/no-vscode-postmessage": "error",
 			},
 		},
 	},

eslint-rules/no-vscode-postmessage.js

@@ -0,0 +1,61 @@
+const { ESLintUtils } = require("@typescript-eslint/utils")
+const path = require("path")
+
+const createRule = ESLintUtils.RuleCreator((name) => `https://cline.bot/eslint-rules/${name}`)
+
+module.exports = createRule({
+	name: "no-vscode-postmessage",
+	meta: {
+		type: "problem",
+		docs: {
+			description: "Ban vscode.postMessage() calls in favor of gRPC service clients, except in grpc-client-base.ts",
+			recommended: "error",
+		},
+		messages: {
+			useGrpcClient:
+				"Use gRPC service clients instead of vscode.postMessage().\n" +
+				"Example: AccountServiceClient.methodName(RequestType.create({...})) instead of vscode.postMessage({type: '...'}).\n" +
+				"Found: {{code}}",
+		},
+		schema: [],
+	},
+	defaultOptions: [],
+
+	create(context) {
+		// Check if current file is grpc-client-base.ts (exception case)
+		const filename = context.filename
+		const isGrpcClientBase = path.basename(filename) === "grpc-client-base.ts"
+
+		return {
+			// Detect vscode.postMessage calls
+			"CallExpression[callee.type='MemberExpression']"(node) {
+				// Skip if this is grpc-client-base.ts
+				if (isGrpcClientBase) {
+					return
+				}
+
+				const callee = node.callee
+
+				// Check for vscode.postMessage pattern
+				if (
+					callee.object &&
+					callee.object.type === "Identifier" &&
+					callee.object.name === "vscode" &&
+					callee.property &&
+					callee.property.name === "postMessage"
+				) {
+					const sourceCode = context.sourceCode
+					const callText = sourceCode.getText(node).trim()
+
+					context.report({
+						node,
+						messageId: "useGrpcClient",
+						data: {
+							code: callText,
+						},
+					})
+				}
+			},
+		}
+	},
+})

evals/.gitignore

@@ -2,5 +2,23 @@ repositories
 
 results/evals.db
 
+diff-edits/cases/
+diff-edits/results/
+
+# Environment variables
+.env
+
+# backwards compatible
 diff_editing/test_cases/
-diff_editing/test_outputs/
+diff_editing/test_outputs/
+
+*.db
+*.db-wal
+*.db-shm
+
+.cache
+
+# Python bytecode cache
+*__pycache__/
+
+diff-edits/cases.zip

evals/README.md

@@ -17,6 +17,7 @@ The evaluation system consists of two main components:
 
 1. **Test Server**: Enhanced HTTP server in `src/services/test/TestServer.ts` that provides detailed task results
 2. **CLI Tool**: Command-line interface in `evals/cli/` for orchestrating evaluations
+3. **Diff Edit Benchmark**: Separate command using the CLI tool that runs a comprehensive diff editing benchmark suite on real world cases, along with a streamlit dashboard displaying the results. For more details, see the [Diff Edit Benchmark README](./diff-edits/README.md). Make sure you add a `evals/diff-edits/cases` folder with all the conversation jsons. 
 
 ## Directory Structure
 
@@ -40,6 +41,12 @@ cline-repo/
 │   │   │   └── utils/                # Utility functions
 │   │   ├── package.json
 │   │   └── tsconfig.json
+│   ├── diff-edits/                   # Diff editing evaluation suite
+│   │   ├── cases/                    # Test case JSON files
+│   │   ├── results/                  # Evaluation results
+│   │   ├── diff-apply/               # Diff application logic
+│   │   ├── parsing/                  # Assistant message parsing
+│   │   └── prompts/                  # System prompts
 │   ├── repositories/                 # Cloned benchmark repositories
 │   │   ├── exercism/                 # Modified Exercism (from pashpashpash/evals)
 │   │   ├── swe-bench/                # SWE-Bench repository
@@ -148,6 +155,192 @@ Freelance-style programming tasks from the SWELancer benchmark.
 
 Multi-file software engineering tasks from the Multi-SWE-Bench repository.
 
+## Diff Edit Evaluations
+
+The Cline Evaluation System includes a specialized suite for evaluating how well models can make precise edits to files using the `replace_in_file` tool.
+
+### Overview
+
+Diff edit evaluations test a model's ability to:
+
+1. Understand file content and identify specific sections to modify
+2. Generate correct SEARCH/REPLACE blocks for targeted edits
+3. Successfully apply changes without introducing errors
+
+### Directory Structure
+
+```
+diff-edits/
+├── cases/                  # Test case JSON files
+├── results/                # Evaluation results
+├── ClineWrapper.ts         # Wrapper for model interaction
+├── TestRunner.ts           # Main test execution logic
+├── types.ts                # Type definitions
+├── diff-apply/             # Diff application logic
+├── parsing/                # Assistant message parsing
+└── prompts/                # System prompts
+```
+
+### Creating Test Cases
+
+Test cases are defined as JSON files in the `diff-edits/cases/` directory. Each test case should include:
+
+```json
+{
+  "test_id": "example_test_1",
+  "messages": [
+    {
+      "role": "user",
+      "text": "Please fix the bug in this code...",
+      "images": []
+    },
+    {
+      "role": "assistant",
+      "text": "I'll help you fix that bug..."
+    }
+  ],
+  "file_contents": "// Original file content here\nfunction example() {\n  // Code with bug\n}",
+  "file_path": "src/example.js",
+  "system_prompt_details": {
+    "mcp_string": "",
+    "cwd_value": "/path/to/working/directory",
+    "browser_use": false,
+    "width": 900,
+    "height": 600,
+    "os_value": "macOS",
+    "shell_value": "/bin/zsh",
+    "home_value": "/Users/username",
+    "user_custom_instructions": ""
+  },
+  "original_diff_edit_tool_call_message": ""
+}
+```
+
+### Running Diff Edit Evaluations
+
+#### Single Model Evaluation
+
+```bash
+cd evals/cli
+node dist/index.js run-diff-eval --model-ids "anthropic/claude-3-5-sonnet-20241022"
+```
+
+#### Multi-Model Evaluation
+
+Compare multiple models in a single evaluation run:
+
+```bash
+# Compare Claude and Grok models
+node dist/index.js run-diff-eval \
+  --model-ids "anthropic/claude-3-5-sonnet-20241022,x-ai/grok-beta" \
+  --max-cases 10 \
+  --valid-attempts-per-case 3 \
+  --verbose
+
+# Compare multiple Claude variants
+node dist/index.js run-diff-eval \
+  --model-ids "anthropic/claude-3-5-sonnet-20241022,anthropic/claude-3-5-haiku-20241022,anthropic/claude-3-opus-20240229" \
+  --max-cases 5 \
+  --valid-attempts-per-case 2 \
+  --parallel
+```
+
+#### Options
+
+- `--model-ids`: Comma-separated list of model IDs to evaluate (required)
+- `--system-prompt-name`: System prompt to use (default: "basicSystemPrompt")
+- `--valid-attempts-per-case`: Number of attempts per test case per model (default: 1)
+- `--max-cases`: Maximum number of test cases to run (default: all available)
+- `--parsing-function`: Function to parse assistant messages (default: "parseAssistantMessageV2")
+- `--diff-edit-function`: Function to apply diffs (default: "constructNewFileContentV2")
+- `--test-path`: Path to test cases (default: diff-edits/cases)
+- `--thinking-budget`: Tokens allocated for thinking (default: 0)
+- `--parallel`: Run tests in parallel (flag)
+- `--replay`: Use pre-recorded LLM output (flag)
+- `--verbose`: Enable detailed logging (flag)
+
+#### Examples
+
+```bash
+# Quick test with 2 models, 4 cases, 2 attempts each
+node dist/index.js run-diff-eval \
+  --model-ids "anthropic/claude-3-5-sonnet-20241022,x-ai/grok-beta" \
+  --max-cases 4 \
+  --valid-attempts-per-case 2 \
+  --verbose
+
+# Comprehensive evaluation with parallel execution
+node dist/index.js run-diff-eval \
+  --model-ids "anthropic/claude-3-5-sonnet-20241022,anthropic/claude-3-5-haiku-20241022" \
+  --system-prompt-name claude4SystemPrompt \
+  --valid-attempts-per-case 5 \
+  --max-cases 20 \
+  --parallel \
+  --verbose
+```
+
+### Database Storage & Analytics
+
+All evaluation results are automatically stored in a SQLite database (`diff-edits/evals.db`) for advanced analytics and comparison. The database includes:
+
+- **System Prompts**: Versioned system prompt content with hashing for deduplication
+- **Processing Functions**: Versioned parsing and diff-edit function configurations
+- **Files**: Original and edited file content with content-based hashing
+- **Runs**: Evaluation run metadata and configuration
+- **Cases**: Individual test case information with context tokens
+- **Results**: Detailed results with timing, cost, and success metrics
+
+### Interactive Dashboard
+
+Launch the Streamlit dashboard to visualize and analyze evaluation results:
+
+```bash
+cd diff-edits/dashboard
+streamlit run app.py
+```
+
+The dashboard provides:
+
+- **Model Performance Comparison**: Side-by-side comparison of success rates, latency, and costs
+- **Interactive Charts**: Success rate trends, latency vs cost analysis, and performance metrics
+- **Detailed Drill-Down**: Individual result analysis with file content viewing
+- **Run Selection**: Browse and compare different evaluation runs
+- **Real-time Updates**: Automatically refreshes with new evaluation data
+
+#### Dashboard Features
+
+1. **Hero Section**: Overview of current run with key metrics
+2. **Model Cards**: Performance cards with grades and detailed metrics
+3. **Comparison Charts**: Interactive Plotly charts for visual analysis
+4. **Result Explorer**: Detailed view of individual test results including:
+   - Original and edited file content
+   - Raw model output
+   - Parsed tool calls
+   - Timing and cost metrics
+   - Error analysis
+
+#### Quick Start Dashboard
+
+```bash
+# Run a quick evaluation
+node cli/dist/index.js run-diff-eval \
+  --model-ids "anthropic/claude-3-5-sonnet-20241022,x-ai/grok-beta" \
+  --max-cases 4 \
+  --valid-attempts-per-case 2 \
+  --verbose
+
+# Launch dashboard to view results
+cd diff-edits/dashboard && streamlit run app.py
+```
+
+### Legacy Results
+
+For backward compatibility, results are also saved as JSON files in the `diff-edits/results/` directory. The JSON results include:
+- Success/failure status
+- Extracted tool calls
+- Diff edit content
+- Token usage and cost metrics
+
 ## Metrics
 
 The evaluation system collects the following metrics:

evals/cli/package.json

@@ -1,39 +0,0 @@
-{
-	"name": "cline-evaluation-cli",
-	"version": "0.1.0",
-	"description": "CLI tool for orchestrating Cline evaluations across multiple benchmarks",
-	"main": "dist/index.js",
-	"scripts": {
-		"build": "tsc",
-		"start": "node dist/index.js",
-		"dev": "ts-node src/index.ts",
-		"test": "echo \"Error: no test specified\" && exit 1"
-	},
-	"keywords": [
-		"cline",
-		"evaluation",
-		"benchmark"
-	],
-	"author": "",
-	"license": "MIT",
-	"dependencies": {
-		"better-sqlite3": "^11.10.0",
-		"chalk": "^4.1.2",
-		"commander": "^9.4.1",
-		"execa": "^5.1.1",
-		"node-fetch": "^2.7.0",
-		"ora": "^5.4.1",
-		"sqlite": "^4.1.2",
-		"uuid": "^9.0.0",
-		"yargs": "^17.6.2"
-	},
-	"devDependencies": {
-		"@types/better-sqlite3": "^7.6.3",
-		"@types/node": "^18.11.18",
-		"@types/node-fetch": "^2.6.12",
-		"@types/uuid": "^9.0.0",
-		"@types/yargs": "^17.0.19",
-		"ts-node": "^10.9.1",
-		"typescript": "^4.9.4"
-	}
-}

evals/cli/src/commands/runDiffEval.ts

@@ -3,9 +3,9 @@ import chalk from "chalk"
 import path from "path"
 
 interface RunDiffEvalOptions {
-	modelId: string
+	modelIds: string
 	systemPromptName: string
-	numberOfRuns: number
+	validAttemptsPerCase: number
 	parsingFunction: string
 	diffEditFunction: string
 	thinkingBudget: number
@@ -14,22 +14,25 @@ interface RunDiffEvalOptions {
 	testPath: string
 	outputPath: string
 	replay: boolean
+	replayRunId?: string
+	diffApplyFile?: string
+	maxCases?: number
 }
 
 export async function runDiffEvalHandler(options: RunDiffEvalOptions) {
 	console.log(chalk.blue("Starting diff editing evaluation..."))
 
 	// Resolve the path to the TestRunner.ts script relative to the current file
-	const scriptPath = path.resolve(__dirname, "../../../diff_editing/TestRunner.ts")
+	const scriptPath = path.resolve(__dirname, "../../../diff-edits/TestRunner.ts")
 
 	// Construct the arguments array for the execa call
 	const args = [
-		"--model-id",
-		options.modelId,
+		"--model-ids",
+		options.modelIds,
 		"--system-prompt-name",
 		options.systemPromptName,
-		"--number-of-runs",
-		String(options.numberOfRuns),
+		"--valid-attempts-per-case",
+		String(options.validAttemptsPerCase),
 		"--parsing-function",
 		options.parsingFunction,
 		"--diff-edit-function",
@@ -55,16 +58,28 @@ export async function runDiffEvalHandler(options: RunDiffEvalOptions) {
 		args.push("--replay")
 	}
 
+	if (options.replayRunId) {
+		args.push("--replay-run-id", options.replayRunId)
+	}
+
+	if (options.diffApplyFile) {
+		args.push("--diff-apply-file", options.diffApplyFile)
+	}
+
 	if (options.verbose) {
 		args.push("--verbose")
 	}
 
+	if (options.maxCases) {
+		args.push("--max-cases", String(options.maxCases))
+	}
+
 	try {
 		console.log(chalk.gray(`Executing: npx tsx ${scriptPath} ${args.join(" ")}`))
 
 		// Execute the script as a child process
 		// We use 'inherit' to stream the stdout/stderr directly to the user's terminal
-		const subprocess = execa("npx", ["tsx", scriptPath, ...args], {
+		const subprocess = execa("npx", ["tsx", "--tsconfig", path.resolve(__dirname, "../../../tsconfig.json"), scriptPath, ...args], {
 			stdio: "inherit",
 		})
 

evals/cli/src/index.ts

@@ -84,22 +84,25 @@ program
 	.description("Run the diff editing evaluation suite")
 	.option("--test-path <path>", "Path to the directory containing test case JSON files")
 	.option("--output-path <path>", "Path to the directory to save the test output JSON files")
-	.option("--model-id <model_id>", "The model ID to use for the test")
+	.option("--model-ids <model_ids>", "Comma-separated list of model IDs to test")
 	.option("--system-prompt-name <name>", "The name of the system prompt to use", "basicSystemPrompt")
-	.option("-n, --number-of-runs <number>", "Number of times to run each test case", "1")
+	.option("-n, --valid-attempts-per-case <number>", "Number of valid attempts per test case per model (will retry until this many valid attempts are collected)", "1")
+	.option("--max-cases <number>", "Maximum number of test cases to run (limits total cases loaded)")
 	.option("--parsing-function <name>", "The parsing function to use", "parseAssistantMessageV2")
 	.option("--diff-edit-function <name>", "The diff editing function to use", "constructNewFileContentV2")
 	.option("--thinking-budget <tokens>", "Set the thinking tokens budget", "0")
 	.option("--parallel", "Run tests in parallel", false)
 	.option("--replay", "Run evaluation from a pre-recorded LLM output, skipping the API call", false)
+	.option("--replay-run-id <run_id>", "The ID of the run to replay from the database")
+	.option("--diff-apply-file <filename>", "The name of the diff apply file to use for the replay")
 	.option("-v, --verbose", "Enable verbose logging", false)
 	.action(async (options) => {
 		try {
-			// The logic here simplifies slightly
 			const fullOptions = {
 				...options,
-				numberOfRuns: parseInt(options.numberOfRuns, 10),
+				validAttemptsPerCase: parseInt(options.validAttemptsPerCase, 10),
 				thinkingBudget: parseInt(options.thinkingBudget, 10),
+				maxCases: options.maxCases ? parseInt(options.maxCases, 10) : undefined,
 			}
 			await runDiffEvalHandler(fullOptions)
 		} catch (error) {

evals/diff-edits/ClineWrapper.ts

@@ -9,6 +9,7 @@ import {
 	AssistantMessageContent,
 } from "./parsing/parse-assistant-message-06-06-25" // "../../src/core/assistant-message"
 import { constructNewFileContent as constructNewFileContentV1, constructNewFileContentV2 } from "./diff-apply/diff-06-06-25"
+import { constructNewFileContent as constructNewFileContentV2_1 } from "./diff-apply/diff-06-23-25"
 import { constructNewFileContent as constructNewFileContentV3 } from "../../src/core/assistant-message/diff" // this defaults to the new v1 when called
 
 type ParseAssistantMessageFn = (message: string) => AssistantMessageContent[]
@@ -21,6 +22,8 @@ const parsingFunctions: Record<string, ParseAssistantMessageFn> = {
 }
 
 const diffEditingFunctions: Record<string, ConstructNewFileContentFn> = {
+	"diff-06-06-25": constructNewFileContentV2,
+	"diff-06-23-25": constructNewFileContentV2_1,
 	constructNewFileContentV1: constructNewFileContentV1,
 	constructNewFileContentV2: constructNewFileContentV2,
 	constructNewFileContentV3: constructNewFileContentV3, // position invariant diff
@@ -39,16 +42,22 @@ interface StreamResult {
 		cacheReadTokens: number
 		totalCost: number
 	}
+	timing?: {
+		timeToFirstTokenMs: number
+		timeToFirstEditMs?: number
+		totalRoundTripMs: number
+	}
 }
 
 /**
- * Process the stream and return full response
+ * Process the stream and return full response with timing data
  */
 async function processStream(
 	handler: OpenRouterHandler,
 	systemPrompt: string,
 	messages: Anthropic.Messages.MessageParam[],
 ): Promise<StreamResult> {
+	const startTime = Date.now()
 	const stream = handler.createMessage(systemPrompt, messages)
 
 	let assistantMessage = ""
@@ -58,12 +67,21 @@ async function processStream(
 	let cacheWriteTokens = 0
 	let cacheReadTokens = 0
 	let totalCost = 0
+	
+	// Timing tracking
+	let timeToFirstTokenMs: number | null = null
+	let timeToFirstEditMs: number | null = null
 
 	for await (const chunk of stream) {
 		if (!chunk) {
 			continue
 		}
 
+		// Capture time to first token (any chunk type)
+		if (timeToFirstTokenMs === null) {
+			timeToFirstTokenMs = Date.now() - startTime
+		}
+
 		switch (chunk.type) {
 			case "usage":
 				inputTokens += chunk.inputTokens
@@ -79,10 +97,25 @@ async function processStream(
 				break
 			case "text":
 				assistantMessage += chunk.text
+				
+				// Try to detect first tool call by parsing accumulated message
+				if (timeToFirstEditMs === null) {
+					try {
+						const parsed = parseAssistantMessageV2(assistantMessage)
+						const hasToolCall = parsed.some(block => block.type === "tool_use")
+						if (hasToolCall) {
+							timeToFirstEditMs = Date.now() - startTime
+						}
+					} catch {
+						// Parsing failed, continue accumulating
+					}
+				}
 				break
 		}
 	}
 
+	const totalRoundTripMs = Date.now() - startTime
+
 	return {
 		assistantMessage,
 		reasoningMessage,
@@ -93,6 +126,11 @@ async function processStream(
 			cacheReadTokens,
 			totalCost,
 		},
+		timing: {
+			timeToFirstTokenMs: timeToFirstTokenMs || 0,
+			timeToFirstEditMs: timeToFirstEditMs || undefined,
+			totalRoundTripMs,
+		},
 	}
 }
 
@@ -116,6 +154,7 @@ export async function runSingleEvaluation(input: TestInput): Promise<TestResult>
 			diffEditFunction,
 			thinkingBudgetTokens,
 			originalDiffEditToolCallMessage,
+			diffApplyFile,
 		} = input
 
 		const requiredParams = {
@@ -141,7 +180,7 @@ export async function runSingleEvaluation(input: TestInput): Promise<TestResult>
 		}
 
 		const parseAssistantMessage = parsingFunctions[parsingFunction]
-		const constructNewFileContent = diffEditingFunctions[diffEditFunction]
+		const constructNewFileContent = diffEditingFunctions[diffApplyFile || diffEditFunction]
 
 		if (!parseAssistantMessage || !constructNewFileContent) {
 			return {
@@ -245,7 +284,22 @@ export async function runSingleEvaluation(input: TestInput): Promise<TestResult>
 		}
 
 		// check that we are editing the correct file path
+		console.log(`Expected file path: "${originalFilePath}"`);
+		console.log(`Actual file path used: "${diffToolPath}"`);
 		if (diffToolPath !== originalFilePath) {
+			console.log(`❌ File path mismatch detected!`);
+			// Enhanced logging:
+			if (streamResult?.assistantMessage) {
+				console.log(`   Full model output (assistantMessage):`);
+				console.log(`   -----------------------------------------`);
+				console.log(`   ${streamResult.assistantMessage}`);
+				console.log(`   -----------------------------------------`);
+			}
+			if (toolCall) {
+				console.log(`   Parsed tool call that caused mismatch:`);
+				console.log(`   ${JSON.stringify(toolCall, null, 2)}`);
+				console.log(`   -----------------------------------------`);
+			}
 			return {
 				success: false,
 				streamResult: streamResult,

evals/diff-edits/README.md

@@ -0,0 +1,84 @@
+# A Note on Cline's Diff Evaluation Setup
+
+Hey there, this note explains what we're doing with Cline's diff evaluation (evals) system. It's all about checking how well various AI models (which users connect to Cline via their own API keys), prompts, and diffing tools can handle file changes.
+
+## What We're Trying to Figure Out
+
+The main idea here is to figure out which AI models (configured by users) are best at making `replace_in_file` tool calls that work correctly. This helps us understand model capabilities and also speeds up our own experiments with prompts and diffing algorithms to make Cline better over time. We want to know a few key things.
+
+First, can the model create diffs, which are just sets of SEARCH and REPLACE blocks, that apply cleanly to a file? This is what we call `diffEditSuccess`.
+
+Second, how do different LLMs, like Claude or Grok, stack up against each other when they try to make these diff edits? We use a standard set of real-world test cases for this.
+
+Third, do different system prompts, say our `basicSystemPrompt` versus the `claude4SystemPrompt`, change how well a model does at diff editing?
+
+Fourth, we're also looking at different ways to apply the diffs themselves. We have a few algorithms like `constructNewFileContentV1`, `V2`, and `V3`, and we want to see which ones are more robust when fed model-generated diffs.
+
+Fifth, we track how fast the model starts making an edit. The `timeToFirstEditMs` metric gives us a hint about how quickly a user would see changes happening in their editor.
+
+And finally, we keep an eye on how many tokens are used and what it costs for each model and each try. This helps us compare how efficient they are.
+
+Right now, these evals are mostly about whether the diff *applies* correctly. That means, do the SEARCH blocks find a match, and can the REPLACE blocks be put in without an error? We're not yet deeply analyzing if the change is valid code or matches what the user *wanted* semantically. That's a problem for another day, and will require a lot more scaffolding.
+
+## How We Run These Tests
+
+Two prerequisites:
+
+1. Make sure you have an `evals/.env` file with `OPENROUTER_API_KEY=<your-openrouter-key>` 
+
+2. Make sure you add a `evals/diff-edits/cases` folder with all the conversation jsons prior to running this. 
+
+
+Our testing strategy is based on replaying situations from actual user sessions where diff edits were tried.
+
+It starts with our test cases. Each one is a JSON file in `./cases` that has the conversation history that led to a diff edit, the original file content and its path, and the info needed to rebuild the system prompt from that original session.
+
+Then, for every test run, we set up a specific configuration. This includes which LLM we're testing, which system prompt it gets, which function we use to parse the model's raw output, and which function we use to actually apply the diff. Here's the command I've been using:
+
+```bash
+npm run diff-eval -- --model-ids "anthropic/claude-3-5-sonnet-20241022,x-ai/grok-3-beta" --max-cases 4 --valid-attempts-per-case 2 --verbose --parallel
+```
+
+This will build the eval script, run it, and then open the streamlit dashboard to show the results.
+
+The `TestRunner.ts` script is the main coordinator. For each test case and setup, `ClineWrapper.ts` takes over and sends the conversation and system prompt to the LLM. We then watch the model's response as it streams in and parse it to find any tool calls.
+
+We're specifically looking for the model to make a single `replace_in_file` tool call. Multiple edits in one tool call are allowed, and recorded (in case you want to filter results by number of edits in a single tool call and compare success rate for that slice across different models/system prompts/etc). If it does, and it's for the correct file, we grab the diff content it produced. Then, the chosen diff application algorithm tries to apply that diff to the original file. We record whether this worked or not as `diffEditSuccess`.
+
+We record a bunch of data for every attempt into a database. This includes details about the model and prompt, token counts, costs, the raw output from the model, the parsed tool calls, whether it succeeded or failed, any error messages, and timing info. For a detailed explanation of the database schema, see [database.md](./database.md).
+
+A big part of this is how we handle "valid attempts," which I'll explain next.
+
+## Keeping it Fair with "Valid Attempts"
+
+LLMs can be unpredictable. If we replay an old scenario, a new model, or even the same model later, might do something completely different than what happened originally. It might call another tool or ask a question instead of trying a diff edit.
+
+Since we really want to test the *diff editing* part, we need a way to make sure we're comparing fairly. That's why we have this idea of "valid attempts."
+
+An attempt is "valid" for this benchmark if the model actually tries to do what we're interested in. This means two things. One, it must call the `replace_in_file` tool. Two, it must target the *same file path* that was targeted in the original recorded conversation for that test case.
+
+If the model does something else, like calling a different tool or picking the wrong file, we don't count that attempt against its diff editing score. Instead, we consider it an "invalid attempt" for *this specific benchmark* and simply re-run that test case with that model. We keep doing this until we've collected a set number of these "valid attempts."
+
+For example, if we ask for 5 valid attempts per test case, the system will keep re-rolling for that case until the model has tried to edit the correct file using the `replace_in_file` tool 5 times. Only then do we look at how many of those 5 valid attempts actually resulted in a successful diff application (`diffEditSuccess`).
+
+This way, if we're comparing two models and one gets a 10% success rate on its valid diff edit attempts, and another gets 90%, we have a much clearer picture of their actual diff-generating capabilities. It avoids muddying the waters with attempts where the model didn't even try to perform the specific action we're evaluating. This approach helps us isolate and measure the diff-editing skill more directly, despite the non-deterministic nature of these models.
+
+## Replays
+
+You can also use the replay argument to replay a previous benchmark run. This is super useful for iterating on our diffing algorithms without having to re-run expensive and time-consuming LLM calls.
+
+When you run an evaluation, every detail is stored in the database—including the raw, unmodified output from the model. The replay feature takes advantage of this by pulling that raw output and feeding it into a *different* diffing algorithm. This lets you isolate the performance of the diffing logic itself. We can see if a new algorithm is better at applying the exact same set of diffs that a model generated in a previous run.
+
+This process is blazingly fast and free, as it completely bypasses the need to make new API calls. It ensures a true apples-to-apples comparison between diffing strategies, since the model's output—the "ground truth" for the evaluation—remains identical.
+
+Here’s an example of how you would replay a previous run with a new diffing algorithm:
+
+```shell
+cd evals && npm run diff-eval -- --replay-run-id 9902189e-63a8-4210-a4fc-fe59e2eaf2c2 --diff-apply-file diff-06-23-25 --verbose
+```
+
+In this command:
+-   `--replay-run-id` specifies the original run we want to use as our ground truth.
+-   `--diff-apply-file` tells the script to use the new diffing logic from the `diff-06-23-25.ts` file.
+
+The script will then create a new run in the database that mirrors the original, but with the results of applying the new diffing algorithm. This allows for a direct comparison in the dashboard, helping us quickly see which of our diffing strategies is the most robust.

evals/diff-edits/dashboard/.streamlit/config.toml

@@ -0,0 +1,8 @@
+[theme]
+base="dark"
+
+[browser]
+gatherUsageStats = false
+
+[server]
+headless = true

evals/diff-edits/dashboard/README.md

@@ -0,0 +1,159 @@
+# 🚀 The Sickest Diff Edits Evaluation Dashboard Ever!
+
+A beautiful, modern Streamlit dashboard for visualizing and analyzing diff editing evaluation results with deep drill-down capabilities.
+
+## ✨ Features
+
+### 🎯 **Smart Model Comparison**
+- **Latest Run Focus**: Automatically loads and displays your most recent evaluation run
+- **Beautiful Performance Cards**: Each model gets a stunning card with performance grades (A+ to C)
+- **Best Performer Highlighting**: The top model gets special styling and a trophy 🏆
+- **Interactive Charts**: Success rate comparisons and latency vs cost analysis
+
+### 🔍 **Deep Drill-Down Analysis**
+- **Individual Result Inspection**: Click any model to see detailed results
+- **Side-by-Side File Views**: See original file content with line numbers
+- **Parsed Tool Call Analysis**: View exactly what the model tried to do
+- **Error Analysis**: Detailed error information for failed attempts
+- **Success Metrics**: Line changes, edit counts, and timing breakdowns
+
+### 🎨 **Aesthetic Design**
+- **Modern UI**: Custom CSS with Inter font, gradients, and shadows
+- **Responsive Layout**: Looks great on any screen size
+- **Color-Coded Performance**: Green for excellent, yellow for good, red for poor
+- **Smooth Animations**: Hover effects and transitions
+- **Professional Styling**: Clean, modern design that looks amazing
+
+### 📊 **Comprehensive Metrics**
+- **Success Rates**: Color-coded percentages with performance grades
+- **Timing Analysis**: First token, first edit, and round trip times
+- **Cost Tracking**: Per-result and total cost analysis
+- **Token Metrics**: Context tokens and completion tokens
+- **Edit Statistics**: Number of edits, lines added/deleted
+
+## 🚀 Quick Start
+
+1. **Install dependencies**:
+   ```bash
+   cd diff-edits/dashboard
+   pip install -r requirements.txt
+   ```
+
+2. **Launch the dashboard**:
+   ```bash
+   streamlit run app.py
+   ```
+   
+   Or use the convenient launch script:
+   ```bash
+   ./launch.sh
+   ```
+
+3. **Open your browser** to http://localhost:8501
+
+## 🎯 Dashboard Sections
+
+### **Hero Section**
+- Beautiful gradient header with run information
+- Key metrics overview (models tested, total results, success rate, cost)
+
+### **Model Performance Cards**
+- Each model displayed as a beautiful card
+- Large success rate display with color coding
+- Performance grade badges (A+, A, B+, B, C+, C)
+- Key metrics: latency, cost, results count, first token time
+- "Drill Down" button for detailed analysis
+
+### **Performance Analytics**
+- Interactive bar chart showing success rates
+- Scatter plot of latency vs cost with bubble sizes
+- Hover details and zoom capabilities
+
+### **Detailed Analysis (Drill-Down)**
+- Model-specific success rate, latency, and cost metrics
+- Individual result selector with status icons
+- Tabbed interface for different views:
+
+#### 📄 **File & Edits Tab**
+- **Side-by-side view**: Original file content with line numbers
+- **Edit analysis**: Success/failure status with detailed metrics
+- **Error display**: Clear error information for failed attempts
+- **Success metrics**: Lines added/deleted, number of edits
+- **Parsed tool calls**: JSON view of what the model attempted
+
+#### 🤖 **Raw Output Tab**
+- Complete raw model output in a code viewer
+- Monospace font for easy reading
+
+#### 🔧 **Parsed Tool Call Tab**
+- Pretty-printed JSON of parsed tool calls
+- Diff block visualization for replace_in_file calls
+- Error handling for malformed JSON
+
+#### 📊 **Metrics Tab**
+- Detailed timing metrics (first token, first edit, round trip)
+- Token and cost information
+- Context size and completion tokens
+
+## 🛠 **Technical Features**
+
+### **Smart Data Loading**
+- Automatic latest run detection
+- Efficient SQL queries with proper JOINs
+- Streamlit caching for performance
+- Error handling for missing data
+
+### **Interactive Navigation**
+- Session state management for drill-down views
+- Back button to return to overview
+- Smooth transitions between views
+
+### **Beautiful Styling**
+- Custom CSS with Google Fonts (Inter)
+- Gradient backgrounds and shadows
+- Hover effects and animations
+- Color-coded performance indicators
+- Professional card-based layout
+
+### **Responsive Design**
+- Works on desktop, tablet, and mobile
+- Flexible column layouts
+- Scalable text and metrics
+
+## 🎨 **Design Philosophy**
+
+This dashboard follows modern design principles:
+- **Clarity**: Information is easy to find and understand
+- **Beauty**: Visually appealing with professional styling
+- **Functionality**: Deep drill-down capabilities for detailed analysis
+- **Performance**: Fast loading with efficient data queries
+- **Usability**: Intuitive navigation and clear visual hierarchy
+
+## 📊 **Data Visualization**
+
+- **Plotly Charts**: Interactive, professional-looking visualizations
+- **Color Coding**: Consistent color scheme for performance levels
+- **Performance Badges**: A+ to C grading system
+- **Status Icons**: ✅ for success, ❌ for failure
+- **Metric Cards**: Clean, card-based metric display
+
+## 🔧 **Customization**
+
+The dashboard is highly customizable:
+- **CSS Styling**: Easy to modify colors, fonts, and layouts
+- **Performance Grades**: Adjustable thresholds for A/B/C grades
+- **Metrics Display**: Add or remove metrics as needed
+- **Chart Types**: Easily swap chart types or add new visualizations
+
+## 🚀 **Future Enhancements**
+
+Potential additions:
+- **Historical Trends**: Compare performance across multiple runs
+- **Export Functionality**: Download results as CSV/PDF
+- **Real-time Updates**: Auto-refresh for ongoing evaluations
+- **Custom Filters**: Filter by date range, model type, etc.
+- **Comparison Mode**: Side-by-side model comparisons
+
+---
+
+**This is the sickest eval dashboard ever!** 🔥 It combines beautiful design with powerful analysis capabilities, making it easy to understand model performance at a glance while providing deep drill-down capabilities for detailed investigation.

evals/diff-edits/dashboard/app.py

@@ -0,0 +1,957 @@
+import streamlit as st
+import sqlite3
+import pandas as pd
+import plotly.express as px
+import plotly.graph_objects as go
+from plotly.subplots import make_subplots
+import numpy as np
+from datetime import datetime
+import os
+import json
+import difflib
+# import mimetypes # No longer needed here if guess_language_from_filepath handles it
+from utils import get_database_connection, guess_language_from_filepath # Import from utils
+
+# Page config
+st.set_page_config(
+    page_title="Diff Edits Evaluation Dashboard",
+    page_icon="📊",
+    layout="wide",
+    initial_sidebar_state="expanded"
+)
+
+# Custom CSS for beautiful styling
+st.markdown("""
+<style>
+    /* Import Google Fonts */
+    @import url('https://fonts.googleapis.com/css2?family=Azeret+Mono:wght@400;700&display=swap');
+    
+    /* Global Styles */
+    .main {
+        font-family: 'Azeret Mono', monospace;
+    }
+    
+    /* Hero Section */
+    .hero-container {
+        background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
+        padding: 2rem;
+        border-radius: 15px;
+        margin-bottom: 2rem;
+        color: white;
+        text-align: center;
+    }
+    
+    .hero-title {
+        font-size: 3rem;
+        font-weight: 700;
+        margin-bottom: 0.5rem;
+        text-shadow: 2px 2px 4px rgba(0,0,0,0.3);
+    }
+    
+    .hero-subtitle {
+        font-size: 1.2rem;
+        font-weight: 300;
+        opacity: 0.9;
+    }
+    
+    /* Model Performance Cards */
+    .model-card {
+        background: white;
+        border-radius: 15px;
+        padding: 1.5rem;
+        margin: 1rem 0;
+        box-shadow: 0 8px 32px rgba(0,0,0,0.1);
+        border: 1px solid rgba(255,255,255,0.2);
+        transition: transform 0.3s ease, box-shadow 0.3s ease;
+    }
+    
+    .model-card:hover {
+        transform: translateY(-5px);
+        box-shadow: 0 12px 40px rgba(0,0,0,0.15);
+    }
+    
+    .model-card.best-performer {
+        border: 2px solid #00D4AA;
+        background: linear-gradient(135deg, #f0fdf4 0%, #ecfdf5 100%);
+    }
+    
+    .model-name {
+        font-size: 1.5rem;
+        font-weight: 600;
+        margin-bottom: 1rem;
+        color: #1f2937;
+    }
+    
+    .success-rate {
+        font-size: 3rem;
+        font-weight: 700;
+        margin-bottom: 0.5rem;
+    }
+    
+    .success-rate.excellent { color: #10b981; }
+    .success-rate.good { color: #f59e0b; }
+    .success-rate.poor { color: #ef4444; }
+    
+    .metric-row {
+        display: flex;
+        justify-content: space-between;
+        margin: 0.5rem 0;
+        padding: 0.5rem;
+        background: rgba(0,0,0,0.02);
+        border-radius: 8px;
+    }
+    
+    .metric-label {
+        font-weight: 500;
+        color: #6b7280;
+    }
+    
+    .metric-value {
+        font-weight: 600;
+        color: #1f2937;
+    }
+    
+    /* Performance Badge */
+    .performance-badge {
+        display: inline-block;
+        padding: 0.25rem 0.75rem;
+        border-radius: 20px;
+        font-weight: 600;
+        font-size: 0.875rem;
+        margin-left: 1rem;
+    }
+    
+    .badge-a { background: #10b981; color: white; }
+    .badge-b { background: #f59e0b; color: white; }
+    .badge-c { background: #ef4444; color: white; }
+    
+    /* Comparison Charts */
+    .chart-container {
+        background: white;
+        border-radius: 15px;
+        padding: 1.5rem;
+        margin: 1rem 0;
+        box-shadow: 0 4px 20px rgba(0,0,0,0.08);
+    }
+    
+    /* Result Detail Modal */
+    .result-detail {
+        background: white;
+        border-radius: 15px;
+        padding: 2rem;
+        margin: 1rem 0;
+        box-shadow: 0 8px 32px rgba(0,0,0,0.1);
+    }
+    
+    .file-viewer {
+        background: #f8fafc;
+        border: 1px solid #e2e8f0;
+        border-radius: 8px;
+        padding: 1rem;
+        font-family: 'Monaco', 'Menlo', 'Ubuntu Mono', monospace;
+        font-size: 0.875rem;
+        line-height: 1.5;
+        overflow-x: auto;
+    }
+    
+    .diff-added {
+        background-color: #dcfce7;
+        color: #166534;
+    }
+    
+    .diff-removed {
+        background-color: #fef2f2;
+        color: #dc2626;
+    }
+    
+    .error-display {
+        background: #fef2f2;
+        border: 1px solid #fecaca;
+        border-radius: 8px;
+        padding: 1rem;
+        color: #dc2626;
+        font-family: monospace;
+    }
+    
+    /* Sidebar Styling */
+    .sidebar .sidebar-content {
+        background: linear-gradient(180deg, #f8fafc 0%, #f1f5f9 100%);
+    }
+    
+    /* Custom Metrics */
+    .custom-metric {
+        text-align: center;
+        padding: 1rem;
+        background: white;
+        border-radius: 10px;
+        box-shadow: 0 2px 10px rgba(0,0,0,0.05);
+        margin: 0.5rem 0;
+    }
+    
+    .custom-metric-value {
+        font-size: 2rem;
+        font-weight: 700;
+        color: #1f2937;
+    }
+    
+    .custom-metric-label {
+        font-size: 0.875rem;
+        color: #6b7280;
+        font-weight: 500;
+        margin-top: 0.25rem;
+    }
+</style>
+""", unsafe_allow_html=True)
+
+# Enhanced data loading functions
+@st.cache_data
+def load_all_runs():
+    """Load all evaluation runs"""
+    conn = get_database_connection()
+    
+    query = """
+    SELECT run_id, description, created_at, system_prompt_hash
+    FROM runs 
+    ORDER BY created_at DESC
+    """
+    
+    return pd.read_sql_query(query, conn)
+
+@st.cache_data
+def load_run_comparison(run_id):
+    """Load a specific run with model comparison data"""
+    conn = get_database_connection()
+    
+    # Get the run details
+    run_query = f"""
+    SELECT run_id, description, created_at, system_prompt_hash
+    FROM runs 
+    WHERE run_id = '{run_id}'
+    """
+    run_data = pd.read_sql_query(run_query, conn)
+    
+    if run_data.empty:
+        return None, None
+    
+    # Get model performance for this run
+    model_perf_query = f"""
+    SELECT 
+        res.model_id,
+        COUNT(*) as total_results,
+        AVG(CASE WHEN res.succeeded THEN 1.0 ELSE 0.0 END) as success_rate,
+        AVG(res.cost_usd) as avg_cost,
+        SUM(res.cost_usd) as total_cost,
+        AVG(res.time_to_first_token_ms) as avg_first_token_ms,
+        AVG(res.time_to_first_edit_ms) as avg_first_edit_ms,
+        AVG(res.time_round_trip_ms) as avg_round_trip_ms,
+        AVG(res.completion_tokens) as avg_completion_tokens,
+        AVG(res.num_edits) as avg_num_edits,
+        MIN(res.time_round_trip_ms) as min_round_trip_ms,
+        MAX(res.time_round_trip_ms) as max_round_trip_ms
+    FROM results res
+    JOIN cases c ON res.case_id = c.case_id
+    WHERE c.run_id = '{run_id}'
+      AND (res.error_enum NOT IN (1, 6, 7) OR res.error_enum IS NULL)  -- Exclude: no_tool_calls, wrong_tool_call, wrong_file_edited
+    GROUP BY res.model_id
+    ORDER BY success_rate DESC, avg_round_trip_ms ASC
+    """
+    
+    model_performance = pd.read_sql_query(model_perf_query, conn)
+    
+    return run_data.iloc[0], model_performance
+
+@st.cache_data
+def load_latest_run_comparison():
+    """Load the latest run with model comparison data"""
+    conn = get_database_connection()
+    
+    # Get the latest run
+    latest_run_query = """
+    SELECT run_id, description, created_at, system_prompt_hash
+    FROM runs 
+    ORDER BY created_at DESC 
+    LIMIT 1
+    """
+    latest_run = pd.read_sql_query(latest_run_query, conn)
+    
+    if latest_run.empty:
+        return None, None
+    
+    return load_run_comparison(latest_run.iloc[0]['run_id'])
+
+@st.cache_data
+def load_detailed_results(run_id, model_id=None, valid_only=False):
+    """Load detailed results for drill-down analysis"""
+    conn = get_database_connection()
+    
+    where_clause = f"WHERE c.run_id = '{run_id}'"
+    if model_id:
+        where_clause += f" AND res.model_id = '{model_id}'"
+    
+    # Option to filter out invalid attempts
+    if valid_only:
+        where_clause += " AND (res.error_enum NOT IN (1, 6, 7) OR res.error_enum IS NULL)"
+    
+    query = f"""
+    SELECT 
+        res.*,
+        c.task_id,
+        c.description as case_description,
+        c.tokens_in_context,
+        sp.name as system_prompt_name,
+        pf.name as processing_functions_name,
+        orig_f.filepath as original_filepath,
+        orig_f.content as original_file_content,
+        edit_f.filepath as edited_filepath,
+        edit_f.content as edited_file_content
+    FROM results res
+    JOIN cases c ON res.case_id = c.case_id
+    LEFT JOIN system_prompts sp ON c.system_prompt_hash = sp.hash
+    LEFT JOIN processing_functions pf ON res.processing_functions_hash = pf.hash
+    LEFT JOIN files orig_f ON c.file_hash = orig_f.hash
+    LEFT JOIN files edit_f ON res.file_edited_hash = edit_f.hash
+    {where_clause}
+    ORDER BY res.created_at DESC
+    """
+    
+    return pd.read_sql_query(query, conn)
+
+def get_performance_grade(success_rate):
+    """Get performance grade based on success rate"""
+    if success_rate >= 0.9:
+        return "A+", "excellent"
+    elif success_rate >= 0.8:
+        return "A", "excellent"
+    elif success_rate >= 0.7:
+        return "B+", "good"
+    elif success_rate >= 0.6:
+        return "B", "good"
+    elif success_rate >= 0.5:
+        return "C+", "good"
+    else:
+        return "C", "poor"
+
+def render_hero_section(current_run, model_performance):
+    """Render the hero section with key metrics"""
+    run_title = current_run['description'] if current_run['description'] else f"Run {current_run['run_id'][:8]}..."
+    st.markdown(f"""
+    <div class="hero-container">
+        <div class="hero-title">Diff Edit Evaluation Results</div>
+        <div class="hero-subtitle">A comprehensive analysis of model performance on code editing tasks.</div>
+        <div class="hero-subtitle" style="font-size: 0.9rem; margin-top: 10px;">
+            <strong>Current Run:</strong> {run_title} • {current_run['created_at']}
+        </div>
+    </div>
+    """, unsafe_allow_html=True)
+    
+    # Key metrics row
+    col1, col2, col3, col4 = st.columns(4)
+    
+    total_results = model_performance['total_results'].sum()
+    overall_success = model_performance['success_rate'].mean()
+    total_cost = model_performance['total_cost'].sum()
+    avg_latency = model_performance['avg_round_trip_ms'].mean()
+    
+    with col1:
+        st.markdown(f"""
+        <div class="custom-metric">
+            <div class="custom-metric-value">{len(model_performance)}</div>
+            <div class="custom-metric-label">Models Tested</div>
+        </div>
+        """, unsafe_allow_html=True)
+    
+    with col2:
+        st.markdown(f"""
+        <div class="custom-metric">
+            <div class="custom-metric-value">{total_results}</div>
+            <div class="custom-metric-label">Valid Results</div>
+        </div>
+        """, unsafe_allow_html=True)
+    
+    with col3:
+        success_color = "#10b981" if overall_success > 0.8 else "#f59e0b" if overall_success > 0.6 else "#ef4444"
+        st.markdown(f"""
+        <div class="custom-metric">
+            <div class="custom-metric-value" style="color: {success_color}">{overall_success:.1%}</div>
+            <div class="custom-metric-label">Avg Success Rate</div>
+        </div>
+        """, unsafe_allow_html=True)
+    
+    with col4:
+        st.markdown(f"""
+        <div class="custom-metric">
+            <div class="custom-metric-value">${total_cost:.3f}</div>
+            <div class="custom-metric-label">Total Cost</div>
+        </div>
+        """, unsafe_allow_html=True)
+
+def render_model_comparison_cards(model_performance):
+    """Render beautiful model comparison cards"""
+    st.markdown("## Model Leaderboard")
+    
+    # Find best performer
+    best_model = model_performance.iloc[0]['model_id']
+    
+    for idx, model in model_performance.iterrows():
+        is_best = model['model_id'] == best_model
+        grade, grade_class = get_performance_grade(model['success_rate'])
+        
+        # Create a container for each model
+        with st.container():
+            col1, col2 = st.columns([3, 1])
+            
+            with col1:
+                # Use Streamlit's native components instead of raw HTML
+                if is_best:
+                    st.success(f"**{model['model_id']}** - Best Performer")
+                else:
+                    st.info(f"**{model['model_id']}**")
+                
+                # Success rate with color coding
+                success_rate = model['success_rate']
+                if success_rate >= 0.8:
+                    st.success(f"**Success Rate:** {success_rate:.1%} ({grade})")
+                elif success_rate >= 0.6:
+                    st.warning(f"**Success Rate:** {success_rate:.1%} ({grade})")
+                else:
+                    st.error(f"**Success Rate:** {success_rate:.1%} ({grade})")
+                
+                # Metrics in columns
+                metric_col1, metric_col2, metric_col3, metric_col4 = st.columns(4)
+                
+                with metric_col1:
+                    if pd.notna(model['avg_round_trip_ms']):
+                        st.metric("Avg Latency", f"{model['avg_round_trip_ms']:.0f}ms")
+                    else:
+                        st.metric("Avg Latency", "N/A")
+                
+                with metric_col2:
+                    if pd.notna(model['avg_cost']):
+                        st.metric("Avg Cost", f"${model['avg_cost']:.4f}")
+                    else:
+                        st.metric("Avg Cost", "N/A")
+                
+                with metric_col3:
+                    st.metric("Valid Results", f"{model['total_results']}")
+                
+                with metric_col4:
+                    if pd.notna(model['avg_first_token_ms']):
+                        st.metric("First Token", f"{model['avg_first_token_ms']:.0f}ms")
+                    else:
+                        st.metric("First Token", "N/A")
+            
+            with col2:
+                st.write("")  # Add some spacing
+                if st.button(f"Drill Down", key=f"drill_{model['model_id']}", use_container_width=True):
+                    st.session_state.drill_down_model = model['model_id']
+            
+            st.divider()  # Add a divider between models
+
+def render_comparison_charts(model_performance):
+    """Render interactive comparison charts"""
+    st.markdown("## Performance Analysis")
+    
+    col1, col2 = st.columns(2)
+
+    with col1:
+        # Time to First Edit
+        fig_first_edit = px.bar(
+            model_performance,
+            x='model_id',
+            y='avg_first_edit_ms',
+            title="Time to First Edit",
+            labels={'avg_first_edit_ms': 'Time to First Edit (ms)', 'model_id': 'Model'},
+            color='avg_first_edit_ms',
+            color_continuous_scale='bluered',
+            text='avg_first_edit_ms',
+            template='plotly_dark'
+        )
+        fig_first_edit.update_traces(texttemplate='%{text:.0f}ms', textposition='outside')
+        fig_first_edit.update_layout(
+            showlegend=False,
+            plot_bgcolor='rgba(0,0,0,0)',
+            paper_bgcolor='rgba(0,0,0,0)',
+            font=dict(family="Azeret Mono, monospace"),
+            margin=dict(t=50)
+        )
+        st.plotly_chart(fig_first_edit, use_container_width=True)
+
+    with col2:
+        # Latency vs Cost Scatter
+        fig_scatter = px.scatter(
+            model_performance,
+            x='avg_round_trip_ms',
+            y='avg_cost',
+            size='total_results',
+            color='success_rate',
+            hover_name='model_id',
+            title="Latency vs Cost Analysis",
+            labels={
+                'avg_round_trip_ms': 'Avg Round Trip (ms)',
+                'avg_cost': 'Avg Cost ($)',
+                'success_rate': 'Success Rate',
+                'total_results': 'Valid Results'
+            },
+            color_continuous_scale='RdYlGn',
+            template='plotly_dark'
+        )
+        fig_scatter.update_layout(
+            plot_bgcolor='rgba(0,0,0,0)',
+            paper_bgcolor='rgba(0,0,0,0)',
+            font=dict(family="Azeret Mono, monospace")
+        )
+        st.plotly_chart(fig_scatter, use_container_width=True)
+
+def render_detailed_analysis(run_id, model_id):
+    """Render detailed drill-down analysis"""
+    st.markdown(f"## Detailed Analysis: {model_id}")
+    
+    # Load all results (including invalid attempts)
+    detailed_results = load_detailed_results(run_id, model_id)
+    
+    # Also load only valid results for metrics
+    valid_results = load_detailed_results(run_id, model_id, valid_only=True)
+    
+    if detailed_results.empty:
+        st.warning("No detailed results found.")
+        return
+    
+    # Show total vs valid results
+    st.info(f"Showing all {len(detailed_results)} results ({len(valid_results)} valid, {len(detailed_results) - len(valid_results)} invalid)")
+    
+    # Results overview
+    col1, col2, col3 = st.columns(3)
+    
+    with col1:
+        success_count = valid_results['succeeded'].sum()
+        total_count = len(valid_results)
+        st.metric("Success Rate", f"{success_count}/{total_count} ({success_count/total_count:.1%} of valid results)")
+    
+    with col2:
+        avg_latency = detailed_results['time_round_trip_ms'].mean()
+        st.metric("Avg Latency", f"{avg_latency:.0f}ms")
+    
+    with col3:
+        total_cost = detailed_results['cost_usd'].sum()
+        st.metric("Total Cost", f"${total_cost:.4f}")
+    
+    # Interactive results table
+    st.markdown("### 📋 Individual Results")
+    
+    # Add result selector with indicators for valid/invalid attempts
+    result_options = []
+    for idx, row in detailed_results.iterrows():
+        # Check if this is a valid result
+        is_valid = (row['error_enum'] not in [1, 6, 7]) if not pd.isna(row['error_enum']) else True
+        
+        # Create status indicator
+        if is_valid:
+            status = "✅" if row['succeeded'] else "❌"
+        else:
+            status = "⚠️"  # Warning symbol for invalid results
+            
+        # Add validity indicator to the option text
+        validity_text = "" if is_valid else " [INVALID RESULT]"
+        result_options.append(f"{status} {row['task_id']} - {row['time_round_trip_ms']:.0f}ms{validity_text}")
+    
+    selected_result_idx = st.selectbox(
+        "Select a result to analyze:",
+        range(len(result_options)),
+        format_func=lambda x: result_options[x]
+    )
+    
+    if selected_result_idx is not None:
+        render_result_detail(detailed_results.iloc[selected_result_idx])
+
+def render_result_detail(result):
+    """Render detailed view of a single result"""
+    st.markdown("### 🔬 Result Deep Dive")
+    
+    # Check if this is a valid result
+    is_valid = (result['error_enum'] not in [1, 6, 7]) if not pd.isna(result['error_enum']) else True
+    
+    # Show validity warning if needed
+    if not is_valid:
+        st.warning("⚠️ **This is an invalid result** - The model didn't properly call the diff edit tool or edited the wrong file. This result is excluded from success rate calculations.")
+    
+    # Result metadata
+    col1, col2, col3, col4 = st.columns(4)
+    
+    with col1:
+        status_icon = "✅" if result['succeeded'] else "❌"
+        st.markdown(f"**Status:** {status_icon} {'Success' if result['succeeded'] else 'Failed'}")
+    
+    with col2:
+        st.markdown(f"**Task ID:** {result['task_id']}")
+    
+    with col3:
+        st.markdown(f"**Round Trip:** {result['time_round_trip_ms']:.0f}ms")
+    
+    with col4:
+        st.markdown(f"**Cost:** ${result['cost_usd']:.4f}")
+    
+    # Tabbed interface for different views
+    tab1, tab2, tab3, tab4 = st.tabs(["📄 File & Edits", "🤖 Raw Output", "🔧 Parsed Tool Call", "📊 Metrics"])
+    
+    with tab1:
+        render_file_and_edits_view(result)
+    
+    with tab2:
+        render_raw_output_view(result)
+    
+    with tab3:
+        render_parsed_tool_call_view(result)
+    
+    with tab4:
+        render_metrics_view(result)
+
+def render_file_and_edits_view(result):
+    """Render side-by-side file and edits view"""
+    st.markdown("#### 📄 File Content & Edit Analysis")
+    
+    # Check if we have original file content
+    has_original = not pd.isna(result['original_file_content']) and result['original_file_content']
+    has_edited = not pd.isna(result['edited_file_content']) and result['edited_file_content']
+    
+    if not has_original and not has_edited:
+        st.warning("No file content available for this result.")
+        return
+    
+    col1, col2 = st.columns(2)
+    
+    with col1:
+        st.markdown("**Original File:**")
+        if has_original:
+            filepath = result['original_filepath'] if not pd.isna(result['original_filepath']) else 'Unknown file'
+            st.markdown(f"📁 `{filepath}`")
+            
+            # Display full original file content in a scrollable code block
+            with st.expander("View Original File Content", expanded=True):
+                # Prepare content for the copy button (needs JS-specific escaping)
+                raw_content_for_copy = result['original_file_content']
+                # Escape for JavaScript template literal: backticks, backslashes, newlines
+                js_escaped_content = raw_content_for_copy.replace('\\', '\\\\') \
+                                                       .replace('`', '\\`') \
+                                                       .replace('\r\n', '\\n') \
+                                                       .replace('\n', '\\n') \
+                                                       .replace('\r', '\\n')
+
+                unique_suffix = str(result.name if hasattr(result, 'name') else result['task_id']).replace('-', '_').replace('.', '_')
+                button_id = f"copyBtnOriginal_{unique_suffix}"
+                
+                copy_button_html = f"""
+                    <button id="{button_id}" onclick="copyOriginalToClipboard(`{js_escaped_content}`, '{button_id}')" style="margin-bottom: 10px; padding: 5px 10px; border-radius: 5px; border: 1px solid #ccc; cursor: pointer;">Copy Original File</button>
+                    <script>
+                        if (!window.copyOriginalToClipboard) {{
+                            window.copyOriginalToClipboard = async function(text, buttonId) {{
+                                try {{
+                                    await navigator.clipboard.writeText(text);
+                                    const button = document.getElementById(buttonId);
+                                    button.innerText = 'Copied!';
+                                    button.style.backgroundColor = '#d4edda'; // Optional: success feedback
+                                    setTimeout(() => {{ 
+                                        button.innerText = 'Copy Original File'; 
+                                        button.style.backgroundColor = '';
+                                    }}, 2000);
+                                }} catch (err) {{
+                                    console.error('Failed to copy original: ', err);
+                                    const button = document.getElementById(buttonId);
+                                    button.innerText = 'Copy Failed!';
+                                    button.style.backgroundColor = '#f8d7da'; // Optional: error feedback
+                                    setTimeout(() => {{ 
+                                        button.innerText = 'Copy Original File'; 
+                                        button.style.backgroundColor = '';
+                                    }}, 2000);
+                                }}
+                            }}
+                        }}
+                    </script>
+                    """
+                st.components.v1.html(copy_button_html, height=50)
+
+                # Prepare content for st.code (needs actual newlines)
+                content_for_display = result['original_file_content']
+                # Iteratively replace common escaped newline sequences with actual newlines
+                # This handles cases like "\\n" -> "\n" and then "\n" (if it was literally "\n")
+                # Order might matter if there are multiple levels of escaping, but this covers common ones.
+                content_for_display = content_for_display.replace('\\\\r\\\\n', '\r\n').replace('\\\\n', '\n') # Double escaped
+                content_for_display = content_for_display.replace('\\r\\n', '\r\n').replace('\\n', '\n')     # Single escaped
+
+                language = guess_language_from_filepath(filepath)
+                st.code(content_for_display, language=language, line_numbers=False)
+
+        else:
+            st.warning("Original file content not available")
+    
+    with col2:
+        st.markdown("**Edit Analysis:**")
+        
+        if not result['succeeded']:
+            # Show error information
+            st.error("❌ **Edit Failed**")
+            
+            if not pd.isna(result['error_enum']):
+                st.markdown(f"**Error Code:** {result['error_enum']}")
+        else:
+            # Show successful edit information
+            st.success("✅ **Edit Successful**")
+            
+            # Show edit metrics
+            metric_col1, metric_col2, metric_col3 = st.columns(3)
+            
+            with metric_col1:
+                if not pd.isna(result['num_edits']):
+                    st.metric("Edits", int(result['num_edits']))
+            
+            with metric_col2:
+                if not pd.isna(result['num_lines_added']):
+                    st.metric("Added", int(result['num_lines_added']))
+            
+            with metric_col3:
+                if not pd.isna(result['num_lines_deleted']):
+                    st.metric("Deleted", int(result['num_lines_deleted']))
+            
+            # Show edited file if available
+            if has_edited:
+                st.markdown("**Edited File:**")
+                with st.expander("View Edited File Content"):
+                    edited_lines = result['edited_file_content'].split('\n')
+                    for i, line in enumerate(edited_lines[:50], 1):
+                        st.text(f"{i:3d} | {line}")
+                    
+                    if len(edited_lines) > 50:
+                        st.text(f"... ({len(edited_lines) - 50} more lines)")
+        
+        # Show parsed tool call if available
+        if not pd.isna(result['parsed_tool_call_json']):
+            with st.expander("View Parsed Tool Call"):
+                try:
+                    parsed_call = json.loads(result['parsed_tool_call_json'])
+                    st.json(parsed_call)
+                except:
+                    st.text(result['parsed_tool_call_json'])
+
+def render_raw_output_view(result):
+    """Render raw model output"""
+    st.markdown("#### 🤖 Raw Model Output")
+    
+    if pd.isna(result['raw_model_output']) or not result['raw_model_output']:
+        st.warning("No raw output available for this result.")
+        return
+    
+    st.markdown("""
+    <div class="file-viewer">
+    """, unsafe_allow_html=True)
+    
+    st.text(result['raw_model_output'])
+    
+    st.markdown("</div>", unsafe_allow_html=True)
+
+def render_parsed_tool_call_view(result):
+    """Render parsed tool call analysis"""
+    st.markdown("#### 🔧 Parsed Tool Call Analysis")
+    
+    if pd.isna(result['parsed_tool_call_json']) or not result['parsed_tool_call_json']:
+        st.warning("No parsed tool call available for this result.")
+        return
+    
+    try:
+        parsed_call = json.loads(result['parsed_tool_call_json'])
+        
+        # Pretty print the JSON
+        st.json(parsed_call)
+        
+        # If it's a replace_in_file call, show the diff blocks
+        if isinstance(parsed_call, dict) and 'diff' in parsed_call:
+            st.markdown("**Diff Blocks:**")
+            st.code(parsed_call['diff'], language='diff')
+            
+    except json.JSONDecodeError:
+        st.markdown("**Raw Parsed Call (Invalid JSON):**")
+        st.text(result['parsed_tool_call_json'])
+
+def render_metrics_view(result):
+    """Render detailed metrics for the result"""
+    st.markdown("#### 📊 Detailed Metrics")
+    
+    col1, col2 = st.columns(2)
+    
+    with col1:
+        st.markdown("**Timing Metrics:**")
+        if not pd.isna(result['time_to_first_token_ms']):
+            st.metric("Time to First Token", f"{result['time_to_first_token_ms']:.0f}ms")
+        
+        if not pd.isna(result['time_to_first_edit_ms']):
+            st.metric("Time to First Edit", f"{result['time_to_first_edit_ms']:.0f}ms")
+        
+        if not pd.isna(result['time_round_trip_ms']):
+            st.metric("Round Trip Time", f"{result['time_round_trip_ms']:.0f}ms")
+    
+    with col2:
+        st.markdown("**Token & Cost Metrics:**")
+        if not pd.isna(result['completion_tokens']):
+            st.metric("Completion Tokens", int(result['completion_tokens']))
+        
+        if not pd.isna(result['cost_usd']):
+            st.metric("Cost", f"${result['cost_usd']:.4f}")
+        
+        if not pd.isna(result['tokens_in_context']):
+            st.metric("Context Tokens", int(result['tokens_in_context']))
+
+def guess_language_from_filepath(filepath):
+    """Guess the language for syntax highlighting from filepath."""
+    if not filepath or pd.isna(filepath):
+        return None
+    
+    extension_map = {
+        '.py': 'python',
+        '.js': 'javascript',
+        '.ts': 'typescript',
+        '.java': 'java',
+        '.cs': 'csharp',
+        '.cpp': 'cpp',
+        '.c': 'c',
+        '.html': 'html',
+        '.css': 'css',
+        '.json': 'json',
+        '.sql': 'sql',
+        '.md': 'markdown',
+        '.rb': 'ruby',
+        '.php': 'php',
+        '.go': 'go',
+        '.rs': 'rust',
+        '.swift': 'swift',
+        '.kt': 'kotlin',
+        '.sh': 'bash',
+        '.yaml': 'yaml',
+        '.yml': 'yaml',
+        '.xml': 'xml',
+    }
+    
+    _, ext = os.path.splitext(filepath)
+def main():
+    # Add a note about valid attempts
+    st.sidebar.markdown("""
+    ### Note on Metrics
+    Success rates are calculated based on **valid results only**. 
+    
+    Invalid results (where the model didn't call the diff edit tool or edited the wrong file) are excluded from calculations.
+    """)
+    
+    # Initialize session state
+    if 'drill_down_model' not in st.session_state:
+        st.session_state.drill_down_model = None
+    if 'selected_run_id' not in st.session_state:
+        st.session_state.selected_run_id = None
+    
+    # Load all runs for sidebar
+    all_runs = load_all_runs()
+    
+    if all_runs.empty:
+        st.error("No evaluation runs found in the database.")
+        st.stop()
+    
+    # Sidebar for run selection
+    with st.sidebar:
+        st.markdown("## 📊 Evaluation Runs")
+        st.markdown("Select a run to analyze:")
+        
+        # Create run options with nice formatting
+        run_options = []
+        run_ids = []
+        
+        for idx, run in all_runs.iterrows():
+            # Format the run description nicely
+            date_str = run['created_at'][:10]  # Get just the date part
+            time_str = run['created_at'][11:16]  # Get just the time part
+            
+            if run['description']:
+                display_name = f"🚀 {run['description']}"
+            else:
+                display_name = f"📅 Run {run['run_id'][:8]}..."
+            
+            run_options.append(f"{display_name}\n📅 {date_str} {time_str}")
+            run_ids.append(run['run_id'])
+        
+        # Default to latest run if no selection
+        if st.session_state.selected_run_id is None:
+            default_index = 0  # Latest run is first
+            st.session_state.selected_run_id = run_ids[0]
+        else:
+            try:
+                default_index = run_ids.index(st.session_state.selected_run_id)
+            except ValueError:
+                default_index = 0
+                st.session_state.selected_run_id = run_ids[0]
+        
+        selected_run_idx = st.selectbox(
+            "Choose run:",
+            range(len(run_options)),
+            format_func=lambda x: run_options[x],
+            index=default_index,
+            key="run_selector"
+        )
+        
+        # Update selected run if changed
+        if run_ids[selected_run_idx] != st.session_state.selected_run_id:
+            st.session_state.selected_run_id = run_ids[selected_run_idx]
+            st.session_state.drill_down_model = None  # Reset drill down when changing runs
+            st.rerun()
+        
+        # Show run details in sidebar
+        selected_run = all_runs.iloc[selected_run_idx]
+        st.markdown("---")
+        st.markdown("### 📋 Run Details")
+        st.markdown(f"**Run ID:** `{selected_run['run_id'][:12]}...`")
+        st.markdown(f"**Created:** {selected_run['created_at']}")
+        if selected_run['description']:
+            st.markdown(f"**Description:** {selected_run['description']}")
+    
+    # Load data for selected run
+    current_run, model_performance = load_run_comparison(st.session_state.selected_run_id)
+    
+    if current_run is None or model_performance.empty:
+        st.error("No data found for the selected run.")
+        st.stop()
+    
+    # Render main dashboard
+    render_hero_section(current_run, model_performance)
+    
+    # Check if we're in drill-down mode
+    if st.session_state.drill_down_model:
+        col1, col2 = st.columns([1, 4])
+        with col1:
+            if st.button("Back to Overview", use_container_width=True):
+                st.session_state.drill_down_model = None
+                st.rerun()
+        
+        render_detailed_analysis(current_run['run_id'], st.session_state.drill_down_model)
+    else:
+        # Success Rate Comparison
+        fig_success = px.bar(
+            model_performance,
+            x='model_id',
+            y='success_rate',
+            title="Success Rate by Model",
+            labels={'success_rate': 'Success Rate', 'model_id': 'Model'},
+            color='success_rate',
+            color_continuous_scale='RdYlGn',
+            text='success_rate',
+            template='plotly_dark'
+        )
+        fig_success.update_traces(texttemplate='%{text:.1%}', textposition='outside')
+        fig_success.update_layout(
+            showlegend=False,
+            plot_bgcolor='rgba(0,0,0,0)',
+            paper_bgcolor='rgba(0,0,0,0)',
+            font=dict(family="Azeret Mono, monospace"),
+            yaxis_range=[0,1],  # Set y-axis from 0% to 100%
+            margin=dict(t=50)  # Add top margin to prevent clipping
+        )
+        st.plotly_chart(fig_success, use_container_width=True)
+        
+        render_model_comparison_cards(model_performance)
+        render_comparison_charts(model_performance)
+
+if __name__ == "__main__":
+    main()

evals/diff-edits/dashboard/launch.sh

@@ -0,0 +1,33 @@
+#!/bin/bash
+
+# Diff Edits Evaluation Dashboard Launcher
+echo "🚀 Starting Diff Edits Evaluation Dashboard..."
+
+# Check if we're in the right directory
+if [ ! -f "app.py" ]; then
+    echo "❌ Error: app.py not found. Please run this script from the dashboard directory."
+    exit 1
+fi
+
+# Check if database exists
+if [ ! -f "../evals.db" ]; then
+    echo "⚠️  Warning: Database file ../evals.db not found."
+    echo "   Make sure you've run some evaluations first to populate the database."
+    echo "   You can run: node ../cli/dist/index.js run-diff-eval --model-id anthropic/claude-sonnet-4 --max-cases 1"
+    echo ""
+fi
+
+# Check if requirements are installed
+echo "📦 Checking Python dependencies..."
+if ! python -c "import streamlit, plotly, pandas" 2>/dev/null; then
+    echo "📥 Installing required packages..."
+    pip install -r requirements.txt
+fi
+
+echo "🌐 Launching Streamlit dashboard..."
+echo "   Dashboard will open in your browser at http://localhost:8501"
+echo "   Press Ctrl+C to stop the dashboard"
+echo ""
+
+# Launch Streamlit
+streamlit run app.py

evals/diff-edits/dashboard/pages/02_Bad_Cases.py

@@ -0,0 +1,183 @@
+import streamlit as st
+import pandas as pd
+import json
+import os # Need to import os for load_case_raw_data
+from utils import get_database_connection, guess_language_from_filepath # Absolute import
+
+st.set_page_config(
+    page_title="Case Health Inspector",
+    page_icon="🧑‍⚕️",
+    layout="wide"
+)
+
+st.title("Case Health Inspector")
+st.markdown("Identify test cases that are frequently problematic across different models and runs.")
+
+@st.cache_data
+def load_problematic_cases_summary():
+    conn = get_database_connection()
+    query = """
+    WITH case_attempts AS (
+        SELECT
+            c.task_id,
+            c.description AS case_description,
+            f_orig.filepath AS original_filepath, -- Get from files table
+            r.run_id,
+            r.model_id,
+            r.result_id,
+            (CASE WHEN (r.error_enum NOT IN (1, 6, 7) OR r.error_enum IS NULL) THEN 1 ELSE 0 END) AS is_valid_attempt,
+            (CASE WHEN (r.error_enum NOT IN (1, 6, 7) OR r.error_enum IS NULL) THEN r.succeeded ELSE NULL END) AS succeeded_on_valid
+        FROM cases c
+        JOIN results r ON c.case_id = r.case_id
+        LEFT JOIN files f_orig ON c.file_hash = f_orig.hash -- Join to get original filepath
+    ),
+    case_summary AS (
+        SELECT
+            task_id,
+            case_description,
+            original_filepath, -- This is now f_orig.filepath
+            COUNT(DISTINCT run_id) AS num_benchmark_runs,
+            COUNT(result_id) AS total_attempts,
+            SUM(is_valid_attempt) AS total_valid_attempts,
+            SUM(succeeded_on_valid) AS total_successful_valid_attempts
+        FROM case_attempts
+        GROUP BY task_id, case_description, original_filepath -- original_filepath is f_orig.filepath
+    )
+    SELECT
+        task_id,
+        case_description,
+        original_filepath, -- This is f_orig.filepath from case_summary
+        num_benchmark_runs,
+        total_attempts,
+        total_valid_attempts,
+        CAST(total_valid_attempts AS REAL) * 100.0 / total_attempts AS percent_valid_attempts,
+        CASE 
+            WHEN total_valid_attempts > 0 THEN CAST(total_successful_valid_attempts AS REAL) * 100.0 / total_valid_attempts
+            ELSE 0 
+        END AS success_rate_on_valid
+    FROM case_summary
+    ORDER BY percent_valid_attempts ASC, success_rate_on_valid ASC;
+    """
+    df = pd.read_sql_query(query, conn)
+    return df
+
+@st.cache_data
+def load_case_raw_data(task_id):
+    """Loads the original JSON data for a given task_id."""
+    # This assumes test cases are stored in ../cases relative to this script's parent (dashboard)
+    # So, ../../cases from this script's location (pages/02_Bad_Cases.py)
+    # Correct path from this script (pages/02_Bad_Cases.py) to cases/
+    # os.path.dirname(__file__) -> pages
+    # os.path.join(..., '..') -> dashboard
+    # os.path.join(..., '..', '..') -> diff-edits
+    # os.path.join(..., '..', '..', 'cases') -> diff-edits/cases
+    cases_dir = os.path.join(os.path.dirname(__file__), '..', '..', 'cases')
+    
+    # The task_id is usually the filename without .json
+    # However, some task_ids might have suffixes or be different.
+    # We need a robust way to find the file. For now, assume task_id is filename base.
+    # This might need adjustment if task_id format varies significantly from filename.
+    
+    # Try direct match first
+    potential_filename = f"{task_id}.json"
+    filepath = os.path.join(cases_dir, potential_filename)
+
+    if not os.path.exists(filepath):
+        # If direct match fails, list files and try to find one that starts with task_id
+        # This is a simple fallback, might need more robust matching if task_ids are complex
+        try:
+            for f_name in os.listdir(cases_dir):
+                if f_name.startswith(task_id) and f_name.endswith(".json"):
+                    filepath = os.path.join(cases_dir, f_name)
+                    break
+            else: # No break means no file found
+                 return None # File not found
+        except FileNotFoundError:
+            return None # Cases directory itself not found
+
+    if not os.path.exists(filepath): # Check again after potential find
+        return None
+
+    try:
+        with open(filepath, 'r') as f:
+            return json.load(f)
+    except Exception as e:
+        st.error(f"Error loading case file {filepath}: {e}")
+        return None
+
+def render_problematic_cases_page():
+    summary_df = load_problematic_cases_summary()
+
+    if summary_df.empty:
+        st.warning("No case summary data found. Run some evaluations first.")
+        return
+
+    st.markdown("### Cases Overview")
+    st.dataframe(summary_df.style.format({
+        "percent_valid_attempts": "{:.1f}%",
+        "success_rate_on_valid": "{:.1f}%"
+    }), use_container_width=True)
+
+    st.markdown("---")
+    st.markdown("### Case Drill Down")
+    
+    selected_task_id = st.selectbox(
+        "Select a Case ID (task_id) to inspect:",
+        options=[""] + summary_df['task_id'].tolist() # Add a blank option
+    )
+
+    if selected_task_id:
+        case_data = summary_df[summary_df['task_id'] == selected_task_id].iloc[0]
+        st.subheader(f"Details for Case: {case_data['task_id']}")
+        st.markdown(f"**Description:** {case_data['case_description']}")
+        st.markdown(f"**Original Filepath:** `{case_data['original_filepath']}`")
+        
+        raw_json_data = load_case_raw_data(selected_task_id)
+        if raw_json_data:
+            with st.expander("View Raw Case JSON Data", expanded=False):
+                st.json(raw_json_data)
+            
+            if 'file_contents' in raw_json_data and raw_json_data['file_contents']:
+                with st.expander("View Original File Content (from Case JSON)", expanded=True):
+                    # Prepare content for the copy button
+                    raw_content_for_copy = raw_json_data['file_contents']
+                    js_escaped_content = raw_content_for_copy.replace('\\', '\\\\') \
+                                                           .replace('`', '\\`') \
+                                                           .replace('\r\n', '\\n') \
+                                                           .replace('\n', '\\n') \
+                                                           .replace('\r', '\\n')
+                    button_id = f"copyBtnCase_{selected_task_id.replace('-', '_').replace('.', '_')}"
+                    copy_button_html = f"""
+                        <button id="{button_id}" onclick="copyCaseContentToClipboard(`{js_escaped_content}`, '{button_id}')" style="margin-bottom: 10px; padding: 5px 10px; border-radius: 5px; border: 1px solid #ccc; cursor: pointer;">Copy File Content</button>
+                        <script>
+                            if (!window.copyCaseContentToClipboard) {{
+                                window.copyCaseContentToClipboard = async function(text, buttonId) {{
+                                    try {{
+                                        await navigator.clipboard.writeText(text);
+                                        const button = document.getElementById(buttonId);
+                                        button.innerText = 'Copied!';
+                                        setTimeout(() => {{ button.innerText = 'Copy File Content'; }}, 2000);
+                                    }} catch (err) {{ console.error('Failed to copy: ', err); const button = document.getElementById(buttonId); button.innerText = 'Copy Failed!'; setTimeout(() => {{ button.innerText = 'Copy File Content'; }}, 2000); }}
+                                }}
+                            }}
+                        </script>
+                        """
+                    st.components.v1.html(copy_button_html, height=50)
+
+                    # Prepare content for st.code
+                    content_for_display = raw_json_data['file_contents']
+                    content_for_display = content_for_display.replace('\\\\r\\\\n', '\r\n').replace('\\\\n', '\n')
+                    content_for_display = content_for_display.replace('\\r\\n', '\r\n').replace('\\n', '\n')
+                    
+                    language = guess_language_from_filepath(case_data['original_filepath'])
+                    st.code(content_for_display, language=language, line_numbers=False)
+            else:
+                st.warning("Original file content not found in case JSON.")
+        else:
+            st.error(f"Could not load raw JSON data for case: {selected_task_id}")
+        
+        # Placeholder for more detailed stats (per-model performance on this case, error breakdown)
+        st.markdown("*(Further per-model statistics and error breakdowns for this case can be added here.)*")
+
+if __name__ == "__main__":
+    render_problematic_cases_page()

evals/diff-edits/dashboard/requirements.txt

@@ -0,0 +1,4 @@
+streamlit>=1.28.0
+plotly>=5.17.0
+pandas>=2.0.0
+numpy>=1.24.0

evals/diff-edits/dashboard/utils.py

@@ -0,0 +1,51 @@
+import streamlit as st
+import sqlite3
+import pandas as pd
+import os
+
+@st.cache_resource
+def get_database_connection():
+    # Assuming the script is run from the dashboard directory,
+    # evals.db is two levels up from there.
+    # __file__ is utils.py, its dirname is dashboard.
+    # os.path.dirname(__file__) -> dashboard/
+    # os.path.join(..., '..') -> diff-edits/
+    # os.path.join(..., '..', 'evals.db') -> diff-edits/evals.db
+    db_path = os.path.join(os.path.dirname(__file__), '..', 'evals.db')
+    if not os.path.exists(db_path):
+        st.error(f"Database not found. Expected at: {os.path.abspath(db_path)}")
+        st.stop()
+    return sqlite3.connect(db_path, check_same_thread=False)
+
+def guess_language_from_filepath(filepath):
+    """Guess the language for syntax highlighting from filepath."""
+    if not filepath or pd.isna(filepath):
+        return None
+    
+    extension_map = {
+        '.py': 'python',
+        '.js': 'javascript',
+        '.ts': 'typescript',
+        '.java': 'java',
+        '.cs': 'csharp',
+        '.cpp': 'cpp',
+        '.c': 'c',
+        '.html': 'html',
+        '.css': 'css',
+        '.json': 'json',
+        '.sql': 'sql',
+        '.md': 'markdown',
+        '.rb': 'ruby',
+        '.php': 'php',
+        '.go': 'go',
+        '.rs': 'rust',
+        '.swift': 'swift',
+        '.kt': 'kotlin',
+        '.sh': 'bash',
+        '.yaml': 'yaml',
+        '.yml': 'yaml',
+        '.xml': 'xml',
+    }
+    
+    _, ext = os.path.splitext(str(filepath)) # Ensure filepath is string
+    return extension_map.get(ext.lower(), None)

evals/diff-edits/database.md

@@ -0,0 +1,96 @@
+# Diff Edit Evaluation Database Schema
+
+This document provides an overview of the SQLite database schema used for the diff edit evaluation suite. The database is designed to capture every aspect of the evaluation runs in a structured way, allowing for detailed, multi-dimensional analysis and ensuring full reproducibility of our findings.
+
+## Data Model Overview
+
+The database is composed of several interconnected tables that work together to provide a comprehensive picture of each evaluation. The core of the model revolves around `runs`, `cases`, and `results`.
+
+### `runs`
+
+A `run` represents a single, top-level execution of the evaluation script (e.g., one invocation of `npm run diff-eval`). It serves as the main container for a complete benchmark session.
+
+-   **Purpose**: To group all the results from a single benchmark execution, allowing for high-level comparison between different runs over time.
+-   **Key Columns**:
+    -   `run_id`: A unique identifier for the entire run.
+    -   `description`: A human-readable summary of the run's configuration (e.g., which models were tested, how many cases, etc.).
+    -   `system_prompt_hash`: A foreign key that links this run to the specific system prompt that was used, ensuring we can track performance changes based on prompt modifications.
+
+### `cases`
+
+A `case` represents a single test scenario that is presented to a model. It corresponds to one of the JSON files in the `cases/` directory and links that static definition to a specific benchmark `run`.
+
+-   **Purpose**: To track the individual test scenarios within a given run.
+-   **Key Columns**:
+    -   `case_id`: A unique identifier for the case *within* a specific run.
+    -   `run_id`: A foreign key linking back to the parent `run`.
+    -   `task_id`: The original, persistent identifier for the test case (typically from the JSON filename).
+    -   `file_hash`: A foreign key linking to the original, un-edited file content for this case.
+
+### `results`
+
+This is the most granular and important table in the database. A `result` represents the outcome of a single attempt by a specific model on a specific case.
+
+-   **Purpose**: To store the detailed outcome of every single model attempt, providing the raw data for all quantitative and qualitative analysis.
+-   **Key Columns**:
+    -   `result_id`: The primary key for the result.
+    -   `run_id`, `case_id`, `model_id`, `processing_functions_hash`: A set of foreign keys that precisely situate this result within the context of a specific run, case, model, and set of helper functions.
+    -   `succeeded`: A boolean indicating if the generated diff was applied successfully.
+    -   `error_enum`: A numeric code representing the specific type of error if the attempt failed (e.g., `1` for `no_tool_calls`, `7` for `wrong_file_edited`).
+    -   `num_edits`, `num_lines_deleted`, `num_lines_added`: Quantitative metrics about the structure of the generated diff.
+    -   `time_to_first_token_ms`, `time_to_first_edit_ms`, `time_round_trip_ms`: High-precision timing data to measure model latency.
+    -   `cost_usd`, `completion_tokens`: Cost and token usage metrics for efficiency analysis.
+    -   `raw_model_output`, `file_edited_hash`, `parsed_tool_call_json`: The rich, qualitative data. This includes the model's full, raw response and the parsed tool calls, which are invaluable for debugging and understanding the model's reasoning.
+
+---
+
+## Supporting Tables
+
+The following tables store versioned, deduplicated content to ensure data integrity and efficiency.
+
+### `system_prompts`
+
+-   **Purpose**: Stores the versioned content of the system prompts used in evaluations.
+-   **Key Columns**:
+    -   `hash`: A unique hash of the prompt's content, which acts as the primary key. This prevents duplicate storage of the same prompt.
+    -   `name`: A human-readable name for the prompt (e.g., `basicSystemPrompt`, `claude4SystemPrompt`).
+    -   `content`: The full text of the system prompt.
+
+### `processing_functions`
+
+-   **Purpose**: Stores the versioned combinations of parsing and diff-editing functions.
+-   **Key Columns**:
+    -   `hash`: A unique hash of the function combination name.
+    -   `name`: A human-readable name (e.g., `parseV2-diffV2`).
+    -   `parsing_function`: The name of the function used to parse the model's output.
+    -   `diff_edit_function`: The name of the function used to apply the diff.
+
+### `files`
+
+-   **Purpose**: Stores the content of all files involved in the tests, including the original source files and the diffs generated by the models.
+-   **Key Columns**:
+    -   `hash`: A content-based hash of the file, ensuring that identical files are only stored once.
+    -   `filepath`: The original path of the file.
+    -   `content`: The full content of the file.
+
+## The Bigger Picture
+
+This relational schema provides a powerful foundation for sophisticated analysis. It moves beyond simple pass/fail metrics and allows us to explore the nuanced interactions between models, prompts, and the code they operate on. With this database, we can answer critical questions like:
+
+-   "How does prompt engineering affect not just success rate, but also latency and cost?"
+-   "Are certain models more prone to specific types of errors (e.g., hallucinating file paths vs. failing to call a tool)?"
+-   "Which of our internal diffing algorithms is the most robust against a wide range of model-generated edits?"
+
+Ultimately, this data model enables us to move from simply *measuring* performance to truly *understanding* it, providing the insights needed to build more capable and reliable AI engineering systems.
+
+---
+
+## Viewing the Full Schema
+
+To see the most up-to-date and detailed schema for the database, you can use the `sqlite3` command-line tool. From the `evals/diff-edits` directory, run the following command:
+
+```bash
+sqlite3 evals.db .schema
+```
+
+This will print the complete `CREATE TABLE` statements for all tables in the database, providing a definitive reference for the database structure.

evals/diff-edits/database/client.ts

@@ -0,0 +1,135 @@
+import Database from 'better-sqlite3';
+import * as fs from 'fs';
+import * as path from 'path';
+import * as crypto from 'crypto';
+
+export class DatabaseClient {
+  private static instance: DatabaseClient;
+  private db: Database.Database;
+  private dbPath: string;
+
+  private constructor() {
+    // Get database path from environment or use default
+    this.dbPath = process.env.DIFF_EVALS_DB_PATH || path.join(__dirname, '../evals.db');
+    
+    // Ensure directory exists
+    const dbDir = path.dirname(this.dbPath);
+    if (!fs.existsSync(dbDir)) {
+      fs.mkdirSync(dbDir, { recursive: true });
+    }
+
+    // Initialize database connection
+    this.db = new Database(this.dbPath);
+    
+    // Enable WAL mode for concurrent access
+    this.db.pragma('journal_mode = WAL');
+    
+    // Enable foreign key constraints
+    this.db.pragma('foreign_keys = ON');
+    
+    // Initialize schema if needed
+    this.initializeSchema();
+  }
+
+  static getInstance(): DatabaseClient {
+    if (!DatabaseClient.instance) {
+      DatabaseClient.instance = new DatabaseClient();
+    }
+    return DatabaseClient.instance;
+  }
+
+  private initializeSchema(): void {
+    // Check if tables exist by trying to query one of them
+    try {
+      this.db.prepare('SELECT COUNT(*) FROM system_prompts LIMIT 1').get();
+      // If we get here, tables exist
+      return;
+    } catch (error) {
+      // Tables don't exist, create them
+      console.log('Initializing database schema...');
+      this.createTables();
+    }
+  }
+
+  private createTables(): void {
+    const schemaPath = path.join(__dirname, 'schema.sql');
+    const schema = fs.readFileSync(schemaPath, 'utf8');
+    
+    // Execute the entire schema as one block
+    this.db.transaction(() => {
+      this.db.exec(schema);
+    })();
+
+    console.log('Database schema initialized successfully');
+  }
+
+  getDatabase(): Database.Database {
+    return this.db;
+  }
+
+  getDatabasePath(): string {
+    return this.dbPath;
+  }
+
+  // Utility method to generate SHA-256 hash
+  static generateHash(content: string): string {
+    return crypto.createHash('sha256').update(content).digest('hex');
+  }
+
+  // Utility method to generate UUID-like ID
+  static generateId(): string {
+    return crypto.randomUUID();
+  }
+
+  // Transaction wrapper
+  transaction<T>(fn: () => T): T {
+    return this.db.transaction(fn)();
+  }
+
+  // Close database connection (for cleanup)
+  close(): void {
+    if (this.db) {
+      this.db.close();
+    }
+  }
+
+  // Get database info
+  getInfo(): { path: string; size: number; tables: string[] } {
+    const stats = fs.statSync(this.dbPath);
+    const tables = this.db
+      .prepare("SELECT name FROM sqlite_master WHERE type='table' ORDER BY name")
+      .all()
+      .map((row: any) => row.name);
+
+    return {
+      path: this.dbPath,
+      size: stats.size,
+      tables
+    };
+  }
+
+  // Vacuum database (cleanup and optimize)
+  vacuum(): void {
+    this.db.exec('VACUUM');
+  }
+
+  // Get database statistics
+  getStats(): { [tableName: string]: number } {
+    const tables = ['system_prompts', 'processing_functions', 'files', 'runs', 'cases', 'results'];
+    const stats: { [tableName: string]: number } = {};
+
+    for (const table of tables) {
+      try {
+        const result = this.db.prepare(`SELECT COUNT(*) as count FROM ${table}`).get() as { count: number };
+        stats[table] = result.count;
+      } catch (error) {
+        stats[table] = 0;
+      }
+    }
+
+    return stats;
+  }
+}
+
+// Export singleton instance getter
+export const getDatabase = () => DatabaseClient.getInstance();

evals/diff-edits/database/index.ts

@@ -0,0 +1,23 @@
+// Main database module exports
+export { DatabaseClient, getDatabase } from './client';
+export * from './types';
+export * from './operations';
+export * from './queries';
+
+// Re-export commonly used functions for convenience
+export {
+  upsertSystemPrompt,
+  upsertProcessingFunctions,
+  upsertFile,
+  createBenchmarkRun,
+  createCase,
+  insertResult,
+  getRunStats
+} from './operations';
+
+export {
+  getSuccessRatesByModel,
+  getModelComparisons,
+  getDatabaseSummary,
+  getErrorDistribution
+} from './queries';

evals/diff-edits/database/operations.ts

@@ -0,0 +1,348 @@
+import { DatabaseClient } from './client';
+import {
+  SystemPrompt,
+  ProcessingFunctions,
+  FileRecord,
+  BenchmarkRun,
+  Case,
+  Result,
+  CreateSystemPromptInput,
+  CreateProcessingFunctionsInput,
+  CreateFileInput,
+  CreateBenchmarkRunInput,
+  CreateCaseInput,
+  CreateResultInput
+} from './types';
+
+const db = DatabaseClient.getInstance();
+
+// System Prompts Operations
+export async function upsertSystemPrompt(input: CreateSystemPromptInput): Promise<string> {
+  const hash = DatabaseClient.generateHash(input.content);
+  
+  const stmt = db.getDatabase().prepare(`
+    INSERT OR IGNORE INTO system_prompts (hash, name, content)
+    VALUES (?, ?, ?)
+  `);
+  
+  stmt.run(hash, input.name, input.content);
+  return hash;
+}
+
+export async function getSystemPromptByHash(hash: string): Promise<SystemPrompt | null> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT * FROM system_prompts WHERE hash = ?
+  `);
+  
+  const result = stmt.get(hash) as SystemPrompt | undefined;
+  return result || null;
+}
+
+// Processing Functions Operations
+export async function upsertProcessingFunctions(input: CreateProcessingFunctionsInput): Promise<string> {
+  const hash = DatabaseClient.generateHash(input.parsing_function + input.diff_edit_function);
+  
+  const stmt = db.getDatabase().prepare(`
+    INSERT OR IGNORE INTO processing_functions (hash, name, parsing_function, diff_edit_function)
+    VALUES (?, ?, ?, ?)
+  `);
+  
+  stmt.run(hash, input.name, input.parsing_function, input.diff_edit_function);
+  return hash;
+}
+
+export async function getProcessingFunctionsByHash(hash: string): Promise<ProcessingFunctions | null> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT * FROM processing_functions WHERE hash = ?
+  `);
+  
+  const result = stmt.get(hash) as ProcessingFunctions | undefined;
+  return result || null;
+}
+
+// Files Operations
+export async function upsertFile(input: CreateFileInput): Promise<string> {
+  const hash = DatabaseClient.generateHash(input.content);
+  
+  const stmt = db.getDatabase().prepare(`
+    INSERT OR IGNORE INTO files (hash, filepath, content, tokens)
+    VALUES (?, ?, ?, ?)
+  `);
+  
+  stmt.run(hash, input.filepath, input.content, input.tokens || null);
+  return hash;
+}
+
+export async function getFileByHash(hash: string): Promise<FileRecord | null> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT * FROM files WHERE hash = ?
+  `);
+  
+  const result = stmt.get(hash) as FileRecord | undefined;
+  return result || null;
+}
+
+// Benchmark Runs Operations
+export async function createBenchmarkRun(input: CreateBenchmarkRunInput): Promise<string> {
+  const runId = DatabaseClient.generateId();
+  
+  const stmt = db.getDatabase().prepare(`
+    INSERT INTO runs (run_id, description, system_prompt_hash)
+    VALUES (?, ?, ?)
+  `);
+  
+  stmt.run(runId, input.description || null, input.system_prompt_hash);
+  return runId;
+}
+
+export async function getBenchmarkRun(runId: string): Promise<BenchmarkRun | null> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT * FROM runs WHERE run_id = ?
+  `);
+  
+  const result = stmt.get(runId) as BenchmarkRun | undefined;
+  return result || null;
+}
+
+export async function getAllBenchmarkRuns(): Promise<BenchmarkRun[]> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT * FROM runs ORDER BY created_at DESC
+  `);
+  
+  return stmt.all() as BenchmarkRun[];
+}
+
+// Cases Operations
+export async function createCase(input: CreateCaseInput): Promise<string> {
+  const caseId = DatabaseClient.generateId();
+  
+  const stmt = db.getDatabase().prepare(`
+    INSERT INTO cases (case_id, run_id, description, system_prompt_hash, task_id, tokens_in_context, file_hash)
+    VALUES (?, ?, ?, ?, ?, ?, ?)
+  `);
+  
+  stmt.run(
+    caseId,
+    input.run_id,
+    input.description,
+    input.system_prompt_hash,
+    input.task_id,
+    input.tokens_in_context,
+    input.file_hash || null
+  );
+  
+  return caseId;
+}
+
+export async function getCasesByRun(runId: string): Promise<Case[]> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT * FROM cases WHERE run_id = ? ORDER BY created_at
+  `);
+  
+  return stmt.all(runId) as Case[];
+}
+
+export async function getCaseById(caseId: string): Promise<Case | null> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT * FROM cases WHERE case_id = ?
+  `);
+  
+  const result = stmt.get(caseId) as Case | undefined;
+  return result || null;
+}
+
+// Results Operations
+export async function insertResult(input: CreateResultInput): Promise<string> {
+  const resultId = DatabaseClient.generateId();
+  
+  const stmt = db.getDatabase().prepare(`
+    INSERT INTO results (
+      result_id, run_id, case_id, model_id, processing_functions_hash,
+      succeeded, error_enum, num_edits, num_lines_deleted, num_lines_added,
+      time_to_first_token_ms, time_to_first_edit_ms, time_round_trip_ms,
+      cost_usd, completion_tokens, raw_model_output, file_edited_hash,
+      parsed_tool_call_json
+    ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+  `);
+  
+  stmt.run(
+    resultId,
+    input.run_id,
+    input.case_id,
+    input.model_id,
+    input.processing_functions_hash,
+    input.succeeded ? 1 : 0, // Convert boolean to integer
+    input.error_enum || null,
+    input.num_edits || null,
+    input.num_lines_deleted || null,
+    input.num_lines_added || null,
+    input.time_to_first_token_ms || null,
+    input.time_to_first_edit_ms || null,
+    input.time_round_trip_ms || null,
+    input.cost_usd || null,
+    input.completion_tokens || null,
+    input.raw_model_output || null,
+    input.file_edited_hash || null,
+    input.parsed_tool_call_json || null
+  );
+  
+  return resultId;
+}
+
+export async function getResultsByRun(runId: string): Promise<Result[]> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT * FROM results WHERE run_id = ? ORDER BY created_at
+  `);
+  
+  return stmt.all(runId) as Result[];
+}
+
+export async function getResultsByCase(caseId: string): Promise<Result[]> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT * FROM results WHERE case_id = ? ORDER BY created_at
+  `);
+  
+  return stmt.all(caseId) as Result[];
+}
+
+export async function getResultById(resultId: string): Promise<Result | null> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT * FROM results WHERE result_id = ?
+  `);
+  
+  const result = stmt.get(resultId) as Result | undefined;
+  return result || null;
+}
+
+// Batch operations for performance
+export async function insertResultsBatch(inputs: CreateResultInput[]): Promise<string[]> {
+  const stmt = db.getDatabase().prepare(`
+    INSERT INTO results (
+      result_id, run_id, case_id, model_id, processing_functions_hash,
+      succeeded, error_enum, num_edits, num_lines_deleted, num_lines_added,
+      time_to_first_token_ms, time_to_first_edit_ms, time_round_trip_ms,
+      cost_usd, completion_tokens, raw_model_output, file_edited_hash,
+      parsed_tool_call_json
+    ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+  `);
+  
+  return db.transaction(() => {
+    const resultIds: string[] = [];
+    
+    for (const input of inputs) {
+      const resultId = DatabaseClient.generateId();
+      
+      stmt.run(
+        resultId,
+        input.run_id,
+        input.case_id,
+        input.model_id,
+        input.processing_functions_hash,
+        input.succeeded ? 1 : 0, // Convert boolean to integer
+        input.error_enum || null,
+        input.num_edits || null,
+        input.num_lines_deleted || null,
+        input.num_lines_added || null,
+        input.time_to_first_token_ms || null,
+        input.time_to_first_edit_ms || null,
+        input.time_round_trip_ms || null,
+        input.cost_usd || null,
+        input.completion_tokens || null,
+        input.raw_model_output || null,
+        input.file_edited_hash || null,
+        input.parsed_tool_call_json || null
+      );
+      
+      resultIds.push(resultId);
+    }
+    
+    return resultIds;
+  });
+}
+
+export async function createCasesBatch(inputs: CreateCaseInput[]): Promise<string[]> {
+  const stmt = db.getDatabase().prepare(`
+    INSERT INTO cases (case_id, run_id, description, system_prompt_hash, task_id, tokens_in_context)
+    VALUES (?, ?, ?, ?, ?, ?)
+  `);
+  
+  return db.transaction(() => {
+    const caseIds: string[] = [];
+    
+    for (const input of inputs) {
+      const caseId = DatabaseClient.generateId();
+      
+      stmt.run(
+        caseId,
+        input.run_id,
+        input.description,
+        input.system_prompt_hash,
+        input.task_id,
+        input.tokens_in_context
+      );
+      
+      caseIds.push(caseId);
+    }
+    
+    return caseIds;
+  });
+}
+
+// Utility functions
+export async function getRunStats(runId: string): Promise<{
+  total_cases: number;
+  total_results: number;
+  success_rate: number;
+  avg_cost: number;
+  avg_latency: number;
+}> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT 
+      COUNT(DISTINCT c.case_id) as total_cases,
+      COUNT(r.result_id) as total_results,
+      AVG(CASE WHEN r.succeeded THEN 1.0 ELSE 0.0 END) as success_rate,
+      AVG(r.cost_usd) as avg_cost,
+      AVG(r.time_round_trip_ms) as avg_latency
+    FROM cases c
+    LEFT JOIN results r ON c.case_id = r.case_id
+    WHERE c.run_id = ?
+  `);
+  
+  const result = stmt.get(runId) as any;
+  return {
+    total_cases: result.total_cases || 0,
+    total_results: result.total_results || 0,
+    success_rate: result.success_rate || 0,
+    avg_cost: result.avg_cost || 0,
+    avg_latency: result.avg_latency || 0
+  };
+}
+
+// Count valid attempts for a specific case and model
+export async function getValidAttemptCount(caseId: string, modelId: string): Promise<number> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT COUNT(*) as count
+    FROM results 
+    WHERE case_id = ? 
+      AND model_id = ?
+      AND error_enum NOT IN (1, 6, 7)  -- Exclude: no_tool_calls, wrong_tool_call, wrong_file_edited
+  `);
+  
+  const result = stmt.get(caseId, modelId) as { count: number };
+  return result.count;
+}
+
+// Get valid results for a specific case and model (for analysis)
+export async function getValidResults(caseId: string, modelId: string, limit?: number): Promise<Result[]> {
+  const limitClause = limit ? `LIMIT ${limit}` : '';
+  const stmt = db.getDatabase().prepare(`
+    SELECT * FROM results 
+    WHERE case_id = ? 
+      AND model_id = ?
+      AND error_enum NOT IN (1, 6, 7)  -- Only valid attempts
+    ORDER BY created_at
+    ${limitClause}
+  `);
+  
+  return stmt.all(caseId, modelId) as Result[];
+}

evals/diff-edits/database/queries.ts

@@ -0,0 +1,309 @@
+import { DatabaseClient } from './client';
+import {
+  ModelSuccessRate,
+  ModelLatency,
+  CostAnalysis,
+  ErrorDistribution,
+  FailedCase,
+  PerformanceTrend,
+  ModelComparison
+} from './types';
+
+const db = DatabaseClient.getInstance();
+
+// Performance analysis queries
+export async function getSuccessRatesByModel(): Promise<ModelSuccessRate[]> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT 
+      model_id,
+      COUNT(*) as total_runs,
+      SUM(CASE WHEN succeeded THEN 1 ELSE 0 END) as successful_runs,
+      ROUND(AVG(CASE WHEN succeeded THEN 1.0 ELSE 0.0 END) * 100, 2) as success_rate
+    FROM results
+    WHERE error_enum NOT IN (1, 6, 7) OR error_enum IS NULL  -- Exclude: no_tool_calls, wrong_tool_call, wrong_file_edited
+    GROUP BY model_id
+    ORDER BY success_rate DESC, total_runs DESC
+  `);
+  
+  return stmt.all() as ModelSuccessRate[];
+}
+
+export async function getAverageLatencyByModel(): Promise<ModelLatency[]> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT 
+      model_id,
+      ROUND(AVG(time_to_first_token_ms), 2) as avg_time_to_first_token_ms,
+      ROUND(AVG(time_to_first_edit_ms), 2) as avg_time_to_first_edit_ms,
+      ROUND(AVG(time_round_trip_ms), 2) as avg_time_round_trip_ms
+    FROM results
+    WHERE time_to_first_token_ms IS NOT NULL
+    GROUP BY model_id
+    ORDER BY avg_time_round_trip_ms ASC
+  `);
+  
+  return stmt.all() as ModelLatency[];
+}
+
+export async function getCostAnalysisByRun(): Promise<CostAnalysis[]> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT 
+      run_id,
+      model_id,
+      ROUND(SUM(cost_usd), 4) as total_cost_usd,
+      ROUND(AVG(cost_usd), 4) as avg_cost_per_case,
+      SUM(completion_tokens) as total_completion_tokens
+    FROM results
+    WHERE cost_usd IS NOT NULL
+    GROUP BY run_id, model_id
+    ORDER BY total_cost_usd DESC
+  `);
+  
+  return stmt.all() as CostAnalysis[];
+}
+
+// Error analysis queries
+export async function getErrorDistribution(): Promise<ErrorDistribution[]> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT 
+      error_enum,
+      COUNT(*) as count,
+      ROUND(COUNT(*) * 100.0 / (SELECT COUNT(*) FROM results WHERE succeeded = 0), 2) as percentage
+    FROM results
+    WHERE succeeded = 0 AND error_enum IS NOT NULL
+    GROUP BY error_enum
+    ORDER BY count DESC
+  `);
+  
+  return stmt.all() as ErrorDistribution[];
+}
+
+export async function getFailedCasesByError(errorEnum?: number): Promise<FailedCase[]> {
+  let query = `
+    SELECT 
+      r.case_id,
+      r.model_id,
+      r.error_enum,
+      c.description,
+      r.raw_model_output
+    FROM results r
+    JOIN cases c ON r.case_id = c.case_id
+    WHERE r.succeeded = 0
+  `;
+  
+  const params: any[] = [];
+  if (errorEnum !== undefined) {
+    query += ` AND r.error_enum = ?`;
+    params.push(errorEnum);
+  }
+  
+  query += ` ORDER BY r.created_at DESC LIMIT 100`;
+  
+  const stmt = db.getDatabase().prepare(query);
+  return stmt.all(...params) as FailedCase[];
+}
+
+// Trend analysis queries
+export async function getPerformanceTrends(days: number = 30): Promise<PerformanceTrend[]> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT 
+      DATE(r.created_at) as date,
+      r.model_id,
+      ROUND(AVG(CASE WHEN r.succeeded THEN 1.0 ELSE 0.0 END) * 100, 2) as success_rate,
+      ROUND(AVG(r.time_round_trip_ms), 2) as avg_latency_ms,
+      ROUND(AVG(r.cost_usd), 4) as avg_cost_usd
+    FROM results r
+    WHERE r.created_at >= datetime('now', '-' || ? || ' days')
+      AND (r.error_enum NOT IN (1, 6, 7) OR r.error_enum IS NULL)  -- Exclude: no_tool_calls, wrong_tool_call, wrong_file_edited
+    GROUP BY DATE(r.created_at), r.model_id
+    ORDER BY date DESC, model_id
+  `);
+  
+  return stmt.all(days) as PerformanceTrend[];
+}
+
+export async function getModelComparisons(): Promise<ModelComparison[]> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT 
+      model_id,
+      ROUND(AVG(CASE WHEN succeeded THEN 1.0 ELSE 0.0 END) * 100, 2) as success_rate,
+      ROUND(AVG(time_round_trip_ms), 2) as avg_latency_ms,
+      ROUND(AVG(cost_usd), 4) as avg_cost_usd,
+      COUNT(*) as total_runs
+    FROM results
+    WHERE error_enum NOT IN (1, 6, 7) OR error_enum IS NULL  -- Exclude: no_tool_calls, wrong_tool_call, wrong_file_edited
+    GROUP BY model_id
+    HAVING total_runs >= 10
+    ORDER BY success_rate DESC, avg_latency_ms ASC
+  `);
+  
+  return stmt.all() as ModelComparison[];
+}
+
+// Advanced analysis queries
+export async function getTopPerformingCases(limit: number = 10): Promise<Array<{
+  case_id: string;
+  description: string;
+  success_rate: number;
+  avg_latency_ms: number;
+  total_runs: number;
+}>> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT 
+      c.case_id,
+      c.description,
+      ROUND(AVG(CASE WHEN r.succeeded THEN 1.0 ELSE 0.0 END) * 100, 2) as success_rate,
+      ROUND(AVG(r.time_round_trip_ms), 2) as avg_latency_ms,
+      COUNT(r.result_id) as total_runs
+    FROM cases c
+    JOIN results r ON c.case_id = r.case_id
+    WHERE r.error_enum NOT IN (1, 6, 7) OR r.error_enum IS NULL  -- Exclude: no_tool_calls, wrong_tool_call, wrong_file_edited
+    GROUP BY c.case_id, c.description
+    HAVING total_runs >= 5
+    ORDER BY success_rate DESC, avg_latency_ms ASC
+    LIMIT ?
+  `);
+  
+  return stmt.all(limit) as Array<{
+    case_id: string;
+    description: string;
+    success_rate: number;
+    avg_latency_ms: number;
+    total_runs: number;
+  }>;
+}
+
+export async function getWorstPerformingCases(limit: number = 10): Promise<Array<{
+  case_id: string;
+  description: string;
+  success_rate: number;
+  avg_latency_ms: number;
+  total_runs: number;
+}>> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT 
+      c.case_id,
+      c.description,
+      ROUND(AVG(CASE WHEN r.succeeded THEN 1.0 ELSE 0.0 END) * 100, 2) as success_rate,
+      ROUND(AVG(r.time_round_trip_ms), 2) as avg_latency_ms,
+      COUNT(r.result_id) as total_runs
+    FROM cases c
+    JOIN results r ON c.case_id = r.case_id
+    WHERE r.error_enum NOT IN (1, 6, 7) OR r.error_enum IS NULL  -- Exclude: no_tool_calls, wrong_tool_call, wrong_file_edited
+    GROUP BY c.case_id, c.description
+    HAVING total_runs >= 5
+    ORDER BY success_rate ASC, avg_latency_ms DESC
+    LIMIT ?
+  `);
+  
+  return stmt.all(limit) as Array<{
+    case_id: string;
+    description: string;
+    success_rate: number;
+    avg_latency_ms: number;
+    total_runs: number;
+  }>;
+}
+
+export async function getModelPerformanceByTimeOfDay(): Promise<Array<{
+  model_id: string;
+  hour: number;
+  success_rate: number;
+  avg_latency_ms: number;
+  total_runs: number;
+}>> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT 
+      model_id,
+      CAST(strftime('%H', created_at) AS INTEGER) as hour,
+      ROUND(AVG(CASE WHEN succeeded THEN 1.0 ELSE 0.0 END) * 100, 2) as success_rate,
+      ROUND(AVG(time_round_trip_ms), 2) as avg_latency_ms,
+      COUNT(*) as total_runs
+    FROM results
+    GROUP BY model_id, hour
+    HAVING total_runs >= 5
+    ORDER BY model_id, hour
+  `);
+  
+  return stmt.all() as Array<{
+    model_id: string;
+    hour: number;
+    success_rate: number;
+    avg_latency_ms: number;
+    total_runs: number;
+  }>;
+}
+
+export async function getRunComparison(runId1: string, runId2: string): Promise<{
+  run1: { run_id: string; success_rate: number; avg_latency_ms: number; avg_cost_usd: number; total_cases: number };
+  run2: { run_id: string; success_rate: number; avg_latency_ms: number; avg_cost_usd: number; total_cases: number };
+}> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT 
+      run_id,
+      ROUND(AVG(CASE WHEN succeeded THEN 1.0 ELSE 0.0 END) * 100, 2) as success_rate,
+      ROUND(AVG(time_round_trip_ms), 2) as avg_latency_ms,
+      ROUND(AVG(cost_usd), 4) as avg_cost_usd,
+      COUNT(DISTINCT case_id) as total_cases
+    FROM results
+    WHERE run_id IN (?, ?)
+    GROUP BY run_id
+  `);
+  
+  const results = stmt.all(runId1, runId2) as Array<{
+    run_id: string;
+    success_rate: number;
+    avg_latency_ms: number;
+    avg_cost_usd: number;
+    total_cases: number;
+  }>;
+  
+  const run1 = results.find(r => r.run_id === runId1);
+  const run2 = results.find(r => r.run_id === runId2);
+  
+  if (!run1 || !run2) {
+    throw new Error('One or both runs not found');
+  }
+  
+  return { run1, run2 };
+}
+
+// Summary statistics
+export async function getDatabaseSummary(): Promise<{
+  total_runs: number;
+  total_cases: number;
+  total_results: number;
+  valid_results: number;
+  unique_models: number;
+  overall_success_rate: number;
+  date_range: { earliest: string; latest: string };
+}> {
+  const stmt = db.getDatabase().prepare(`
+    SELECT 
+      (SELECT COUNT(*) FROM runs) as total_runs,
+      (SELECT COUNT(*) FROM cases) as total_cases,
+      (SELECT COUNT(*) FROM results) as total_results,
+      (SELECT COUNT(*) FROM results WHERE error_enum NOT IN (1, 6, 7) OR error_enum IS NULL) as valid_results,
+      (SELECT COUNT(DISTINCT model_id) FROM results) as unique_models,
+      (SELECT ROUND(AVG(CASE WHEN succeeded THEN 1.0 ELSE 0.0 END) * 100, 2) 
+       FROM results 
+       WHERE error_enum NOT IN (1, 6, 7) OR error_enum IS NULL) as overall_success_rate,
+      (SELECT MIN(created_at) FROM results) as earliest,
+      (SELECT MAX(created_at) FROM results) as latest
+    FROM results
+    LIMIT 1
+  `);
+  
+  const result = stmt.get() as any;
+  return {
+    total_runs: result.total_runs || 0,
+    total_cases: result.total_cases || 0,
+    total_results: result.total_results || 0,
+    valid_results: result.valid_results || 0,
+    unique_models: result.unique_models || 0,
+    overall_success_rate: result.overall_success_rate || 0,
+    date_range: {
+      earliest: result.earliest || '',
+      latest: result.latest || ''
+    }
+  };
+}

evals/diff-edits/database/schema.sql

@@ -0,0 +1,78 @@
+PRAGMA foreign_keys = ON;
+
+CREATE TABLE system_prompts (
+    hash TEXT PRIMARY KEY,
+    name TEXT NOT NULL,
+    content TEXT NOT NULL,
+    created_at DATETIME DEFAULT CURRENT_TIMESTAMP
+);
+
+CREATE TABLE processing_functions (
+    hash TEXT PRIMARY KEY,
+    name TEXT NOT NULL,
+    parsing_function TEXT NOT NULL,
+    diff_edit_function TEXT NOT NULL,
+    created_at DATETIME DEFAULT CURRENT_TIMESTAMP
+);
+
+CREATE TABLE files (
+    hash TEXT PRIMARY KEY,
+    filepath TEXT NOT NULL,
+    content TEXT NOT NULL,
+    tokens INTEGER,
+    created_at DATETIME DEFAULT CURRENT_TIMESTAMP
+);
+
+CREATE TABLE runs (
+    run_id TEXT PRIMARY KEY,
+    created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
+    description TEXT,
+    system_prompt_hash TEXT NOT NULL,
+    FOREIGN KEY (system_prompt_hash) REFERENCES system_prompts(hash)
+);
+
+CREATE TABLE cases (
+    case_id TEXT PRIMARY KEY,
+    run_id TEXT NOT NULL,
+    created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
+    description TEXT NOT NULL,
+    system_prompt_hash TEXT NOT NULL,
+    task_id TEXT NOT NULL,
+    tokens_in_context INTEGER,
+    file_hash TEXT,
+    FOREIGN KEY (run_id) REFERENCES runs(run_id),
+    FOREIGN KEY (system_prompt_hash) REFERENCES system_prompts(hash),
+    FOREIGN KEY (file_hash) REFERENCES files(hash)
+);
+
+CREATE TABLE results (
+    result_id TEXT PRIMARY KEY,
+    run_id TEXT NOT NULL,
+    case_id TEXT NOT NULL,
+    model_id TEXT NOT NULL,
+    processing_functions_hash TEXT NOT NULL,
+    succeeded BOOLEAN NOT NULL,
+    error_enum INTEGER,
+    num_edits INTEGER,
+    num_lines_deleted INTEGER,
+    num_lines_added INTEGER,
+    time_to_first_token_ms INTEGER,
+    time_to_first_edit_ms INTEGER,
+    time_round_trip_ms INTEGER,
+    cost_usd REAL,
+    completion_tokens INTEGER,
+    raw_model_output TEXT,
+    file_edited_hash TEXT,
+    parsed_tool_call_json TEXT,
+    created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
+    FOREIGN KEY (run_id) REFERENCES runs(run_id),
+    FOREIGN KEY (case_id) REFERENCES cases(case_id),
+    FOREIGN KEY (processing_functions_hash) REFERENCES processing_functions(hash)
+);
+
+CREATE INDEX idx_results_run_model ON results(run_id, model_id);
+CREATE INDEX idx_results_case_model ON results(case_id, model_id);
+CREATE INDEX idx_results_success ON results(succeeded);
+CREATE INDEX idx_cases_run ON cases(run_id);
+CREATE INDEX idx_results_created_at ON results(created_at);
+CREATE INDEX idx_runs_created_at ON runs(created_at);

evals/diff-edits/database/test.ts

@@ -0,0 +1,53 @@
+// Simple test to verify database functionality
+import { getDatabase } from './client';
+import { upsertSystemPrompt, createBenchmarkRun, getDatabaseSummary } from './index';
+
+async function testDatabase() {
+  console.log('Testing database functionality...');
+  
+  try {
+    // Test database connection
+    const db = getDatabase();
+    console.log('✓ Database connection established');
+    console.log('Database path:', db.getDatabasePath());
+    
+    // Test database info
+    const info = db.getInfo();
+    console.log('✓ Database info:', info);
+    
+    // Test database stats
+    const stats = db.getStats();
+    console.log('✓ Database stats:', stats);
+    
+    // Test system prompt creation
+    const systemPromptHash = await upsertSystemPrompt({
+      name: 'test-prompt',
+      content: 'This is a test system prompt for database verification.'
+    });
+    console.log('✓ System prompt created with hash:', systemPromptHash);
+    
+    // Test benchmark run creation
+    const runId = await createBenchmarkRun({
+      description: 'Test run for database verification',
+      system_prompt_hash: systemPromptHash
+    });
+    console.log('✓ Benchmark run created with ID:', runId);
+    
+    // Test database summary
+    const summary = await getDatabaseSummary();
+    console.log('✓ Database summary:', summary);
+    
+    console.log('\n🎉 All database tests passed!');
+    
+  } catch (error) {
+    console.error('❌ Database test failed:', error);
+    process.exit(1);
+  }
+}
+
+// Run test if this file is executed directly
+if (require.main === module) {
+  testDatabase();
+}
+
+export { testDatabase };

evals/diff-edits/database/types.ts

@@ -0,0 +1,169 @@
+// Database type definitions for diff-edits evaluation system
+
+export interface SystemPrompt {
+  hash: string;
+  name: string;
+  content: string;
+  created_at: string;
+}
+
+export interface ProcessingFunctions {
+  hash: string;
+  name: string;
+  parsing_function: string;
+  diff_edit_function: string;
+  created_at: string;
+}
+
+export interface FileRecord {
+  hash: string;
+  filepath: string;
+  content: string;
+  tokens?: number;
+  created_at: string;
+}
+
+export interface BenchmarkRun {
+  run_id: string;
+  created_at: string;
+  description?: string;
+  system_prompt_hash: string;
+}
+
+export interface Case {
+	case_id: string
+	run_id: string
+	created_at: string
+	description: string
+	system_prompt_hash: string
+	task_id: string
+	tokens_in_context: number
+	file_hash?: string
+}
+
+export interface Result {
+  result_id: string;
+  run_id: string;
+  case_id: string;
+  model_id: string;
+  processing_functions_hash: string;
+  succeeded: boolean;
+  error_enum?: number;
+  num_edits?: number;
+  num_lines_deleted?: number;
+  num_lines_added?: number;
+  time_to_first_token_ms?: number;
+  time_to_first_edit_ms?: number;
+  time_round_trip_ms?: number;
+  cost_usd?: number;
+  completion_tokens?: number;
+  raw_model_output?: string;
+  file_edited_hash?: string;
+  parsed_tool_call_json?: string;
+  created_at: string;
+}
+
+// Input types for creating records
+export interface CreateSystemPromptInput {
+  name: string;
+  content: string;
+}
+
+export interface CreateProcessingFunctionsInput {
+  name: string;
+  parsing_function: string;
+  diff_edit_function: string;
+}
+
+export interface CreateFileInput {
+  filepath: string;
+  content: string;
+  tokens?: number;
+}
+
+export interface CreateBenchmarkRunInput {
+  description?: string;
+  system_prompt_hash: string;
+}
+
+export interface CreateCaseInput {
+  run_id: string;
+  description: string;
+  system_prompt_hash: string;
+  task_id: string;
+  tokens_in_context: number;
+  file_hash?: string;
+}
+
+export interface CreateResultInput {
+  run_id: string;
+  case_id: string;
+  model_id: string;
+  processing_functions_hash: string;
+  succeeded: boolean;
+  error_enum?: number;
+  num_edits?: number;
+  num_lines_deleted?: number;
+  num_lines_added?: number;
+  time_to_first_token_ms?: number;
+  time_to_first_edit_ms?: number;
+  time_round_trip_ms?: number;
+  cost_usd?: number;
+  completion_tokens?: number;
+  raw_model_output?: string;
+  file_edited_hash?: string;
+  parsed_tool_call_json?: string;
+}
+
+// Analysis result types
+export interface ModelSuccessRate {
+  model_id: string;
+  total_runs: number;
+  successful_runs: number;
+  success_rate: number;
+}
+
+export interface ModelLatency {
+  model_id: string;
+  avg_time_to_first_token_ms: number;
+  avg_time_to_first_edit_ms: number;
+  avg_time_round_trip_ms: number;
+}
+
+export interface CostAnalysis {
+  run_id: string;
+  model_id: string;
+  total_cost_usd: number;
+  avg_cost_per_case: number;
+  total_completion_tokens: number;
+}
+
+export interface ErrorDistribution {
+  error_enum: number;
+  count: number;
+  percentage: number;
+}
+
+export interface FailedCase {
+  case_id: string;
+  model_id: string;
+  error_enum: number;
+  description: string;
+  raw_model_output?: string;
+}
+
+export interface PerformanceTrend {
+  date: string;
+  model_id: string;
+  success_rate: number;
+  avg_latency_ms: number;
+  avg_cost_usd: number;
+}
+
+export interface ModelComparison {
+  model_id: string;
+  success_rate: number;
+  avg_latency_ms: number;
+  avg_cost_usd: number;
+  total_runs: number;
+}

evals/diff-edits/diff-apply/diff-06-23-25.ts

@@ -0,0 +1,827 @@
+const SEARCH_BLOCK_START = "------- SEARCH"
+const SEARCH_BLOCK_END = "======="
+const REPLACE_BLOCK_END = "+++++++ REPLACE"
+
+const SEARCH_BLOCK_CHAR = "-"
+const REPLACE_BLOCK_CHAR = "+"
+const LEGACY_SEARCH_BLOCK_CHAR = "<"
+const LEGACY_REPLACE_BLOCK_CHAR = ">"
+
+// Replace the exact string constants with flexible regex patterns
+const SEARCH_BLOCK_START_REGEX = /^[-]{3,} SEARCH$/
+const SEARCH_BLOCK_END_REGEX = /^[=]{3,}$/
+const REPLACE_BLOCK_END_REGEX = /^[+]{3,} REPLACE$/
+const LEGACY_SEARCH_BLOCK_START_REGEX = /^[<]{3,} SEARCH$/
+const LEGACY_REPLACE_BLOCK_END_REGEX = /^[>]{3,} REPLACE$/
+
+// Helper functions to check if a line matches the flexible patterns
+function isSearchBlockStart(line: string): boolean {
+	return SEARCH_BLOCK_START_REGEX.test(line) || LEGACY_SEARCH_BLOCK_START_REGEX.test(line)
+}
+
+function isSearchBlockEnd(line: string): boolean {
+	return SEARCH_BLOCK_END_REGEX.test(line)
+}
+
+function isReplaceBlockEnd(line: string): boolean {
+	return REPLACE_BLOCK_END_REGEX.test(line) || LEGACY_REPLACE_BLOCK_END_REGEX.test(line)
+}
+
+/**
+ * Attempts a line-trimmed fallback match for the given search content in the original content.
+ * It tries to match `searchContent` lines against a block of lines in `originalContent` starting
+ * from `lastProcessedIndex`. Lines are matched by trimming leading/trailing whitespace and ensuring
+ * they are identical afterwards.
+ *
+ * Returns [matchIndexStart, matchIndexEnd] if found, or false if not found.
+ */
+function lineTrimmedFallbackMatch(originalContent: string, searchContent: string, startIndex: number): [number, number] | false {
+	// Split both contents into lines
+	const originalLines = originalContent.split("\n")
+	const searchLines = searchContent.split("\n")
+
+	// Trim trailing empty line if exists (from the trailing \n in searchContent)
+	if (searchLines[searchLines.length - 1] === "") {
+		searchLines.pop()
+	}
+
+	// Find the line number where startIndex falls
+	let startLineNum = 0
+	let currentIndex = 0
+	while (currentIndex < startIndex && startLineNum < originalLines.length) {
+		currentIndex += originalLines[startLineNum].length + 1 // +1 for \n
+		startLineNum++
+	}
+
+	// For each possible starting position in original content
+	for (let i = startLineNum; i <= originalLines.length - searchLines.length; i++) {
+		let matches = true
+
+		// Try to match all search lines from this position
+		for (let j = 0; j < searchLines.length; j++) {
+			const originalTrimmed = originalLines[i + j].trim()
+			const searchTrimmed = searchLines[j].trim()
+
+			if (originalTrimmed !== searchTrimmed) {
+				matches = false
+				break
+			}
+		}
+
+		// If we found a match, calculate the exact character positions
+		if (matches) {
+			// Find start character index
+			let matchStartIndex = 0
+			for (let k = 0; k < i; k++) {
+				matchStartIndex += originalLines[k].length + 1 // +1 for \n
+			}
+
+			// Find end character index
+			let matchEndIndex = matchStartIndex
+			for (let k = 0; k < searchLines.length; k++) {
+				matchEndIndex += originalLines[i + k].length + 1 // +1 for \n
+			}
+
+			return [matchStartIndex, matchEndIndex]
+		}
+	}
+
+	return false
+}
+
+/**
+ * Attempts to match blocks of code by using the first and last lines as anchors.
+ * This is a third-tier fallback strategy that helps match blocks where we can identify
+ * the correct location by matching the beginning and end, even if the exact content
+ * differs slightly.
+ *
+ * The matching strategy:
+ * 1. Only attempts to match blocks of 3 or more lines to avoid false positives
+ * 2. Extracts from the search content:
+ *    - First line as the "start anchor"
+ *    - Last line as the "end anchor"
+ * 3. For each position in the original content:
+ *    - Checks if the next line matches the start anchor
+ *    - If it does, jumps ahead by the search block size
+ *    - Checks if that line matches the end anchor
+ *    - All comparisons are done after trimming whitespace
+ *
+ * This approach is particularly useful for matching blocks of code where:
+ * - The exact content might have minor differences
+ * - The beginning and end of the block are distinctive enough to serve as anchors
+ * - The overall structure (number of lines) remains the same
+ *
+ * @param originalContent - The full content of the original file
+ * @param searchContent - The content we're trying to find in the original file
+ * @param startIndex - The character index in originalContent where to start searching
+ * @returns A tuple of [startIndex, endIndex] if a match is found, false otherwise
+ */
+function blockAnchorFallbackMatch(originalContent: string, searchContent: string, startIndex: number): [number, number] | false {
+	const originalLines = originalContent.split("\n")
+	const searchLines = searchContent.split("\n")
+
+	// Only use this approach for blocks of 3+ lines
+	if (searchLines.length < 3) {
+		return false
+	}
+
+	// Trim trailing empty line if exists
+	if (searchLines[searchLines.length - 1] === "") {
+		searchLines.pop()
+	}
+
+	const firstLineSearch = searchLines[0].trim()
+	const lastLineSearch = searchLines[searchLines.length - 1].trim()
+	const searchBlockSize = searchLines.length
+
+	// Find the line number where startIndex falls
+	let startLineNum = 0
+	let currentIndex = 0
+	while (currentIndex < startIndex && startLineNum < originalLines.length) {
+		currentIndex += originalLines[startLineNum].length + 1
+		startLineNum++
+	}
+
+	// Look for matching start and end anchors
+	for (let i = startLineNum; i <= originalLines.length - searchBlockSize; i++) {
+		// Check if first line matches
+		if (originalLines[i].trim() !== firstLineSearch) {
+			continue
+		}
+
+		// Check if last line matches at the expected position
+		if (originalLines[i + searchBlockSize - 1].trim() !== lastLineSearch) {
+			continue
+		}
+
+		// Calculate exact character positions
+		let matchStartIndex = 0
+		for (let k = 0; k < i; k++) {
+			matchStartIndex += originalLines[k].length + 1
+		}
+
+		let matchEndIndex = matchStartIndex
+		for (let k = 0; k < searchBlockSize; k++) {
+			matchEndIndex += originalLines[i + k].length + 1
+		}
+
+		return [matchStartIndex, matchEndIndex]
+	}
+
+	return false
+}
+
+/**
+ * This function reconstructs the file content by applying a streamed diff (in a
+ * specialized SEARCH/REPLACE block format) to the original file content. It is designed
+ * to handle both incremental updates and the final resulting file after all chunks have
+ * been processed.
+ *
+ * The diff format is a custom structure that uses three markers to define changes:
+ *
+ *   ------- SEARCH
+ *   [Exact content to find in the original file]
+ *   =======
+ *   [Content to replace with]
+ *   +++++++ REPLACE
+ *
+ * Behavior and Assumptions:
+ * 1. The file is processed chunk-by-chunk. Each chunk of `diffContent` may contain
+ *    partial or complete SEARCH/REPLACE blocks. By calling this function with each
+ *    incremental chunk (with `isFinal` indicating the last chunk), the final reconstructed
+ *    file content is produced.
+ *
+ * 2. Matching Strategy (in order of attempt):
+ *    a. Exact Match: First attempts to find the exact SEARCH block text in the original file
+ *    b. Line-Trimmed Match: Falls back to line-by-line comparison ignoring leading/trailing whitespace
+ *    c. Block Anchor Match: For blocks of 3+ lines, tries to match using first/last lines as anchors
+ *    If all matching strategies fail, an error is thrown.
+ *
+ * 3. Empty SEARCH Section:
+ *    - If SEARCH is empty and the original file is empty, this indicates creating a new file
+ *      (pure insertion).
+ *    - If SEARCH is empty and the original file is not empty, this indicates a complete
+ *      file replacement (the entire original content is considered matched and replaced).
+ *
+ * 4. Applying Changes:
+ *    - Before encountering the "=======" marker, lines are accumulated as search content.
+ *    - After "=======" and before ">>>>>>> REPLACE", lines are accumulated as replacement content.
+ *    - Once the block is complete (">>>>>>> REPLACE"), the matched section in the original
+ *      file is replaced with the accumulated replacement lines, and the position in the original
+ *      file is advanced.
+ *
+ * 5. Incremental Output:
+ *    - As soon as the match location is found and we are in the REPLACE section, each new
+ *      replacement line is appended to the result so that partial updates can be viewed
+ *      incrementally.
+ *
+ * 6. Partial Markers:
+ *    - If the final line of the chunk looks like it might be part of a marker but is not one
+ *      of the known markers, it is removed. This prevents incomplete or partial markers
+ *      from corrupting the output.
+ *
+ * 7. Finalization:
+ *    - Once all chunks have been processed (when `isFinal` is true), any remaining original
+ *      content after the last replaced section is appended to the result.
+ *    - Trailing newlines are not forcibly added. The code tries to output exactly what is specified.
+ *
+ * Errors:
+ * - If the search block cannot be matched using any of the available matching strategies,
+ *   an error is thrown.
+ */
+export async function constructNewFileContent(
+	diffContent: string,
+	originalContent: string,
+	isFinal: boolean,
+	version: "v1" | "v2" = "v1",
+): Promise<string> {
+	const constructor = constructNewFileContentVersionMapping[version]
+	if (!constructor) {
+		throw new Error(`Invalid version '${version}' for file content constructor`)
+	}
+	return constructor(diffContent, originalContent, isFinal)
+}
+
+const constructNewFileContentVersionMapping: Record<
+	string,
+	(diffContent: string, originalContent: string, isFinal: boolean) => Promise<string>
+> = {
+	v1: constructNewFileContentV1,
+	v2: constructNewFileContentV2,
+} as const
+
+async function constructNewFileContentV1(diffContent: string, originalContent: string, isFinal: boolean): Promise<string> {
+	let result = ""
+	let lastProcessedIndex = 0
+
+	let currentSearchContent = ""
+	let currentReplaceContent = ""
+	let inSearch = false
+	let inReplace = false
+
+	let searchMatchIndex = -1
+	let searchEndIndex = -1
+
+	// Track all replacements to handle out-of-order edits
+	let replacements: Array<{ start: number; end: number; content: string }> = []
+	let pendingOutOfOrderReplacement = false
+
+	let lines = diffContent.split("\n")
+
+	// If the last line looks like a partial marker but isn't recognized,
+	// remove it because it might be incomplete.
+	const lastLine = lines[lines.length - 1]
+	if (
+		lines.length > 0 &&
+		(lastLine.startsWith(SEARCH_BLOCK_CHAR) ||
+			lastLine.startsWith(LEGACY_SEARCH_BLOCK_CHAR) ||
+			lastLine.startsWith("=") ||
+			lastLine.startsWith(REPLACE_BLOCK_CHAR) ||
+			lastLine.startsWith(LEGACY_REPLACE_BLOCK_CHAR)) &&
+		!isSearchBlockStart(lastLine) &&
+		!isSearchBlockEnd(lastLine) &&
+		!isReplaceBlockEnd(lastLine)
+	) {
+		lines.pop()
+	}
+
+	for (const line of lines) {
+		if (isSearchBlockStart(line)) {
+			inSearch = true
+			currentSearchContent = ""
+			currentReplaceContent = ""
+			continue
+		}
+
+		if (isSearchBlockEnd(line)) {
+			inSearch = false
+			inReplace = true
+
+			// Remove trailing linebreak for adding the === marker
+			// if (currentSearchContent.endsWith("\r\n")) {
+			// 	currentSearchContent = currentSearchContent.slice(0, -2)
+			// } else if (currentSearchContent.endsWith("\n")) {
+			// 	currentSearchContent = currentSearchContent.slice(0, -1)
+			// }
+
+			if (!currentSearchContent) {
+				// Empty search block
+				if (originalContent.length === 0) {
+					// New file scenario: nothing to match, just start inserting
+					searchMatchIndex = 0
+					searchEndIndex = 0
+				} else {
+					// ERROR: Empty search block with non-empty file indicates malformed SEARCH marker
+					throw new Error(
+						"Empty SEARCH block detected with non-empty file. This usually indicates a malformed SEARCH marker.\n" +
+							"Please ensure your SEARCH marker follows the correct format:\n" +
+							"- Use '------- SEARCH' (7+ dashes + space + SEARCH)\n",
+					)
+				}
+			} else {
+				// Add check for inefficient full-file search
+				// if (currentSearchContent.trim() === originalContent.trim()) {
+				// 	throw new Error(
+				// 		"The SEARCH block contains the entire file content. Please either:\n" +
+				// 			"1. Use an empty SEARCH block to replace the entire file, or\n" +
+				// 			"2. Make focused changes to specific parts of the file that need modification.",
+				// 	)
+				// }
+
+				// Exact search match scenario
+				const exactIndex = originalContent.indexOf(currentSearchContent, lastProcessedIndex)
+				if (exactIndex !== -1) {
+					searchMatchIndex = exactIndex
+					searchEndIndex = exactIndex + currentSearchContent.length
+				} else {
+					// Attempt fallback line-trimmed matching
+					const lineMatch = lineTrimmedFallbackMatch(originalContent, currentSearchContent, lastProcessedIndex)
+					if (lineMatch) {
+						;[searchMatchIndex, searchEndIndex] = lineMatch
+					} else {
+						// Try block anchor fallback for larger blocks
+						const blockMatch = blockAnchorFallbackMatch(originalContent, currentSearchContent, lastProcessedIndex)
+						if (blockMatch) {
+							;[searchMatchIndex, searchEndIndex] = blockMatch
+						} else {
+							// Last resort: search the entire file from the beginning
+							const fullFileIndex = originalContent.indexOf(currentSearchContent, 0)
+							if (fullFileIndex !== -1) {
+								// Found in the file - could be out of order
+								searchMatchIndex = fullFileIndex
+								searchEndIndex = fullFileIndex + currentSearchContent.length
+								if (searchMatchIndex < lastProcessedIndex) {
+									pendingOutOfOrderReplacement = true
+								}
+							} else {
+								throw new Error(
+									`The SEARCH block:\n${currentSearchContent.trimEnd()}\n...does not match anything in the file.`,
+								)
+							}
+						}
+					}
+				}
+			}
+
+			// Check if this is an out-of-order replacement
+			if (searchMatchIndex < lastProcessedIndex) {
+				pendingOutOfOrderReplacement = true
+			}
+
+			// For in-order replacements, output everything up to the match location
+			if (!pendingOutOfOrderReplacement) {
+				result += originalContent.slice(lastProcessedIndex, searchMatchIndex)
+			}
+			continue
+		}
+
+		if (isReplaceBlockEnd(line)) {
+			// Finished one replace block
+
+			// Store this replacement
+			replacements.push({
+				start: searchMatchIndex,
+				end: searchEndIndex,
+				content: currentReplaceContent,
+			})
+
+			// If this was an in-order replacement, advance lastProcessedIndex
+			if (!pendingOutOfOrderReplacement) {
+				lastProcessedIndex = searchEndIndex
+			}
+
+			// Reset for next block
+			inSearch = false
+			inReplace = false
+			currentSearchContent = ""
+			currentReplaceContent = ""
+			searchMatchIndex = -1
+			searchEndIndex = -1
+			pendingOutOfOrderReplacement = false
+			continue
+		}
+
+		// Accumulate content for search or replace
+		// (currentReplaceContent is not being used for anything right now since we directly append to result.)
+		// (We artificially add a linebreak since we split on \n at the beginning. In order to not include a trailing linebreak in the final search/result blocks we need to remove it before using them. This allows for partial line matches to be correctly identified.)
+		// NOTE: search/replace blocks must be arranged in the order they appear in the file due to how we build the content using lastProcessedIndex. We also cannot strip the trailing newline since for non-partial lines it would remove the linebreak from the original content. (If we remove end linebreak from search, then we'd also have to remove it from replace but we can't know if it's a partial line or not since the model may be using the line break to indicate the end of the block rather than as part of the search content.) We require the model to output full lines in order for our fallbacks to work as well.
+		if (inSearch) {
+			currentSearchContent += line + "\n"
+		} else if (inReplace) {
+			currentReplaceContent += line + "\n"
+			// Only output replacement lines immediately for in-order replacements
+			if (searchMatchIndex !== -1 && !pendingOutOfOrderReplacement) {
+				result += line + "\n"
+			}
+		}
+	}
+
+	// If this is the final chunk, we need to apply all replacements and build the final result
+	if (isFinal) {
+		// Handle the case where we're still in replace mode when processing ends
+		// and this is the final chunk - treat it as if we encountered the REPLACE marker
+		if (inReplace && searchMatchIndex !== -1) {
+			// Store this replacement
+			replacements.push({
+				start: searchMatchIndex,
+				end: searchEndIndex,
+				content: currentReplaceContent,
+			})
+
+			// If this was an in-order replacement, advance lastProcessedIndex
+			if (!pendingOutOfOrderReplacement) {
+				lastProcessedIndex = searchEndIndex
+			}
+
+			// Reset state
+			inSearch = false
+			inReplace = false
+			currentSearchContent = ""
+			currentReplaceContent = ""
+			searchMatchIndex = -1
+			searchEndIndex = -1
+			pendingOutOfOrderReplacement = false
+		}
+		// end of handling missing replace marker
+
+		// Sort replacements by start position
+		replacements.sort((a, b) => a.start - b.start)
+
+		// Rebuild the entire result by applying all replacements
+		result = ""
+		let currentPos = 0
+
+		for (const replacement of replacements) {
+			// Add original content up to this replacement
+			result += originalContent.slice(currentPos, replacement.start)
+			// Add the replacement content
+			result += replacement.content
+			// Move position to after the replaced section
+			currentPos = replacement.end
+		}
+
+		// Add any remaining original content
+		result += originalContent.slice(currentPos)
+	}
+
+	return result
+}
+
+enum ProcessingState {
+	Idle = 0,
+	StateSearch = 1 << 0,
+	StateReplace = 1 << 1,
+}
+
+class NewFileContentConstructor {
+	private originalContent: string
+	private isFinal: boolean
+	private state: number
+	private pendingNonStandardLines: string[]
+	private result: string
+	private lastProcessedIndex: number
+	private currentSearchContent: string
+	private currentReplaceContent: string
+	private searchMatchIndex: number
+	private searchEndIndex: number
+
+	constructor(originalContent: string, isFinal: boolean) {
+		this.originalContent = originalContent
+		this.isFinal = isFinal
+		this.pendingNonStandardLines = []
+		this.result = ""
+		this.lastProcessedIndex = 0
+		this.state = ProcessingState.Idle
+		this.currentSearchContent = ""
+		this.currentReplaceContent = ""
+		this.searchMatchIndex = -1
+		this.searchEndIndex = -1
+	}
+
+	private resetForNextBlock() {
+		// Reset for next block
+		this.state = ProcessingState.Idle
+		this.currentSearchContent = ""
+		this.currentReplaceContent = ""
+		this.searchMatchIndex = -1
+		this.searchEndIndex = -1
+	}
+
+	private findLastMatchingLineIndex(regx: RegExp, lineLimit: number) {
+		for (let i = lineLimit; i > 0; ) {
+			i--
+			if (this.pendingNonStandardLines[i].match(regx)) {
+				return i
+			}
+		}
+		return -1
+	}
+
+	private updateProcessingState(newState: ProcessingState) {
+		const isValidTransition =
+			(this.state === ProcessingState.Idle && newState === ProcessingState.StateSearch) ||
+			(this.state === ProcessingState.StateSearch && newState === ProcessingState.StateReplace)
+
+		if (!isValidTransition) {
+			throw new Error(
+				`Invalid state transition.\n` +
+					"Valid transitions are:\n" +
+					"- Idle → StateSearch\n" +
+					"- StateSearch → StateReplace",
+			)
+		}
+
+		this.state |= newState
+	}
+
+	private isStateActive(state: ProcessingState): boolean {
+		return (this.state & state) === state
+	}
+
+	private activateReplaceState() {
+		this.updateProcessingState(ProcessingState.StateReplace)
+	}
+
+	private activateSearchState() {
+		this.updateProcessingState(ProcessingState.StateSearch)
+		this.currentSearchContent = ""
+		this.currentReplaceContent = ""
+	}
+
+	private isSearchingActive(): boolean {
+		return this.isStateActive(ProcessingState.StateSearch)
+	}
+
+	private isReplacingActive(): boolean {
+		return this.isStateActive(ProcessingState.StateReplace)
+	}
+
+	private hasPendingNonStandardLines(pendingNonStandardLineLimit: number): boolean {
+		return this.pendingNonStandardLines.length - pendingNonStandardLineLimit < this.pendingNonStandardLines.length
+	}
+
+	public processLine(line: string) {
+		this.internalProcessLine(line, true, this.pendingNonStandardLines.length)
+	}
+
+	public getResult() {
+		// If this is the final chunk, append any remaining original content
+		if (this.isFinal && this.lastProcessedIndex < this.originalContent.length) {
+			this.result += this.originalContent.slice(this.lastProcessedIndex)
+		}
+		if (this.isFinal && this.state !== ProcessingState.Idle) {
+			throw new Error("File processing incomplete - SEARCH/REPLACE operations still active during finalization")
+		}
+		return this.result
+	}
+
+	private internalProcessLine(
+		line: string,
+		canWritependingNonStandardLines: boolean,
+		pendingNonStandardLineLimit: number,
+	): number {
+		let removeLineCount = 0
+		if (isSearchBlockStart(line)) {
+			removeLineCount = this.trimPendingNonStandardTrailingEmptyLines(pendingNonStandardLineLimit)
+			if (removeLineCount > 0) {
+				pendingNonStandardLineLimit = pendingNonStandardLineLimit - removeLineCount
+			}
+			if (this.hasPendingNonStandardLines(pendingNonStandardLineLimit)) {
+				this.tryFixSearchReplaceBlock(pendingNonStandardLineLimit)
+				canWritependingNonStandardLines && (this.pendingNonStandardLines.length = 0)
+			}
+			this.activateSearchState()
+		} else if (isSearchBlockEnd(line)) {
+			// 校验非标内容
+			if (!this.isSearchingActive()) {
+				this.tryFixSearchBlock(pendingNonStandardLineLimit)
+				canWritependingNonStandardLines && (this.pendingNonStandardLines.length = 0)
+			}
+			this.activateReplaceState()
+			this.beforeReplace()
+		} else if (isReplaceBlockEnd(line)) {
+			if (!this.isReplacingActive()) {
+				this.tryFixReplaceBlock(pendingNonStandardLineLimit)
+				canWritependingNonStandardLines && (this.pendingNonStandardLines.length = 0)
+			}
+			this.lastProcessedIndex = this.searchEndIndex
+			this.resetForNextBlock()
+		} else {
+			// Accumulate content for search or replace
+			// (currentReplaceContent is not being used for anything right now since we directly append to result.)
+			// (We artificially add a linebreak since we split on \n at the beginning. In order to not include a trailing linebreak in the final search/result blocks we need to remove it before using them. This allows for partial line matches to be correctly identified.)
+			// NOTE: search/replace blocks must be arranged in the order they appear in the file due to how we build the content using lastProcessedIndex. We also cannot strip the trailing newline since for non-partial lines it would remove the linebreak from the original content. (If we remove end linebreak from search, then we'd also have to remove it from replace but we can't know if it's a partial line or not since the model may be using the line break to indicate the end of the block rather than as part of the search content.) We require the model to output full lines in order for our fallbacks to work as well.
+			if (this.isReplacingActive()) {
+				this.currentReplaceContent += line + "\n"
+				// Output replacement lines immediately if we know the insertion point
+				if (this.searchMatchIndex !== -1) {
+					this.result += line + "\n"
+				}
+			} else if (this.isSearchingActive()) {
+				this.currentSearchContent += line + "\n"
+			} else {
+				let appendToPendingNonStandardLines = canWritependingNonStandardLines
+				if (appendToPendingNonStandardLines) {
+					// 处理非标内容
+					this.pendingNonStandardLines.push(line)
+				}
+			}
+		}
+		return removeLineCount
+	}
+
+	private beforeReplace() {
+		// Remove trailing linebreak for adding the === marker
+		// if (currentSearchContent.endsWith("\r\n")) {
+		// 	currentSearchContent = currentSearchContent.slice(0, -2)
+		// } else if (currentSearchContent.endsWith("\n")) {
+		// 	currentSearchContent = currentSearchContent.slice(0, -1)
+		// }
+
+		if (!this.currentSearchContent) {
+			// Empty search block
+			if (this.originalContent.length === 0) {
+				// New file scenario: nothing to match, just start inserting
+				this.searchMatchIndex = 0
+				this.searchEndIndex = 0
+			} else {
+				// Complete file replacement scenario: treat the entire file as matched
+				this.searchMatchIndex = 0
+				this.searchEndIndex = this.originalContent.length
+			}
+		} else {
+			// Add check for inefficient full-file search
+			// if (currentSearchContent.trim() === originalContent.trim()) {
+			// 	throw new Error(
+			// 		"The SEARCH block contains the entire file content. Please either:\n" +
+			// 			"1. Use an empty SEARCH block to replace the entire file, or\n" +
+			// 			"2. Make focused changes to specific parts of the file that need modification.",
+			// 	)
+			// }
+			// Exact search match scenario
+			const exactIndex = this.originalContent.indexOf(this.currentSearchContent, this.lastProcessedIndex)
+			if (exactIndex !== -1) {
+				this.searchMatchIndex = exactIndex
+				this.searchEndIndex = exactIndex + this.currentSearchContent.length
+			} else {
+				// Attempt fallback line-trimmed matching
+				const lineMatch = lineTrimmedFallbackMatch(
+					this.originalContent,
+					this.currentSearchContent,
+					this.lastProcessedIndex,
+				)
+				if (lineMatch) {
+					;[this.searchMatchIndex, this.searchEndIndex] = lineMatch
+				} else {
+					// Try block anchor fallback for larger blocks
+					const blockMatch = blockAnchorFallbackMatch(
+						this.originalContent,
+						this.currentSearchContent,
+						this.lastProcessedIndex,
+					)
+					if (blockMatch) {
+						;[this.searchMatchIndex, this.searchEndIndex] = blockMatch
+					} else {
+						throw new Error(
+							`The SEARCH block:\n${this.currentSearchContent.trimEnd()}\n...does not match anything in the file.`,
+						)
+					}
+				}
+			}
+		}
+		if (this.searchMatchIndex < this.lastProcessedIndex) {
+			throw new Error(
+				`The SEARCH block:\n${this.currentSearchContent.trimEnd()}\n...matched an incorrect content in the file.`,
+			)
+		}
+		// Output everything up to the match location
+		this.result += this.originalContent.slice(this.lastProcessedIndex, this.searchMatchIndex)
+	}
+
+	private tryFixSearchBlock(lineLimit: number): number {
+		let removeLineCount = 0
+		if (lineLimit < 0) {
+			lineLimit = this.pendingNonStandardLines.length
+		}
+		if (!lineLimit) {
+			throw new Error("Invalid SEARCH/REPLACE block structure - no lines available to process")
+		}
+		let searchTagRegexp = /^([-]{3,}|[<]{3,}) SEARCH$/
+		const searchTagIndex = this.findLastMatchingLineIndex(searchTagRegexp, lineLimit)
+		if (searchTagIndex !== -1) {
+			let fixLines = this.pendingNonStandardLines.slice(searchTagIndex, lineLimit)
+			fixLines[0] = SEARCH_BLOCK_START
+			for (const line of fixLines) {
+				removeLineCount += this.internalProcessLine(line, false, searchTagIndex)
+			}
+		} else {
+			throw new Error(
+				`Invalid REPLACE marker detected - could not find matching SEARCH block starting from line ${searchTagIndex + 1}`,
+			)
+		}
+		return removeLineCount
+	}
+
+	private tryFixReplaceBlock(lineLimit: number): number {
+		let removeLineCount = 0
+		if (lineLimit < 0) {
+			lineLimit = this.pendingNonStandardLines.length
+		}
+		if (!lineLimit) {
+			throw new Error()
+		}
+		let replaceBeginTagRegexp = /^[=]{3,}$/
+		const replaceBeginTagIndex = this.findLastMatchingLineIndex(replaceBeginTagRegexp, lineLimit)
+		if (replaceBeginTagIndex !== -1) {
+			// // 校验非标内容
+			// if (!this.isSearchingActive()) {
+			// 	removeLineCount += this.tryFixSearchBlock(replaceBeginTagIndex)
+			// }
+			let fixLines = this.pendingNonStandardLines.slice(replaceBeginTagIndex - removeLineCount, lineLimit - removeLineCount)
+			fixLines[0] = SEARCH_BLOCK_END
+			for (const line of fixLines) {
+				removeLineCount += this.internalProcessLine(line, false, replaceBeginTagIndex - removeLineCount)
+			}
+		} else {
+			throw new Error(`Malformed REPLACE block - missing valid separator after line ${replaceBeginTagIndex + 1}`)
+		}
+		return removeLineCount
+	}
+
+	private tryFixSearchReplaceBlock(lineLimit: number): number {
+		let removeLineCount = 0
+		if (lineLimit < 0) {
+			lineLimit = this.pendingNonStandardLines.length
+		}
+		if (!lineLimit) {
+			throw new Error()
+		}
+
+		let replaceEndTagRegexp = /^([+]{3,}|[>]{3,}) REPLACE$/
+		const replaceEndTagIndex = this.findLastMatchingLineIndex(replaceEndTagRegexp, lineLimit)
+		const likeReplaceEndTag = replaceEndTagIndex === lineLimit - 1
+		if (likeReplaceEndTag) {
+			// // 校验非标内容
+			// if (!this.isReplacingActive()) {
+			// 	removeLineCount += this.tryFixReplaceBlock(replaceEndTagIndex)
+			// }
+			let fixLines = this.pendingNonStandardLines.slice(replaceEndTagIndex - removeLineCount, lineLimit - removeLineCount)
+			fixLines[fixLines.length - 1] = REPLACE_BLOCK_END
+			for (const line of fixLines) {
+				removeLineCount += this.internalProcessLine(line, false, replaceEndTagIndex - removeLineCount)
+			}
+		} else {
+			throw new Error("Malformed SEARCH/REPLACE block structure: Missing valid closing REPLACE marker")
+		}
+		return removeLineCount
+	}
+
+	/**
+	 * Removes trailing empty lines from the pendingNonStandardLines array
+	 * @param lineLimit - The index to start checking from (exclusive).
+	 *                    Removes empty lines from lineLimit-1 backwards.
+	 * @returns The number of empty lines removed
+	 */
+	private trimPendingNonStandardTrailingEmptyLines(lineLimit: number): number {
+		let removedCount = 0
+		let i = Math.min(lineLimit, this.pendingNonStandardLines.length) - 1
+
+		while (i >= 0 && this.pendingNonStandardLines[i].trim() === "") {
+			this.pendingNonStandardLines.pop()
+			removedCount++
+			i--
+		}
+
+		return removedCount
+	}
+}
+
+export async function constructNewFileContentV2(diffContent: string, originalContent: string, isFinal: boolean): Promise<string> {
+	let newFileContentConstructor = new NewFileContentConstructor(originalContent, isFinal)
+
+	let lines = diffContent.split("\n")
+
+	// If the last line looks like a partial marker but isn't recognized,
+	// remove it because it might be incomplete.
+	const lastLine = lines[lines.length - 1]
+	if (
+		lines.length > 0 &&
+		(lastLine.startsWith(SEARCH_BLOCK_CHAR) ||
+			lastLine.startsWith(LEGACY_SEARCH_BLOCK_CHAR) ||
+			lastLine.startsWith("=") ||
+			lastLine.startsWith(REPLACE_BLOCK_CHAR) ||
+			lastLine.startsWith(LEGACY_REPLACE_BLOCK_CHAR)) &&
+		lastLine !== SEARCH_BLOCK_START &&
+		lastLine !== SEARCH_BLOCK_END &&
+		lastLine !== REPLACE_BLOCK_END
+	) {
+		lines.pop()
+	}
+
+	for (const line of lines) {
+		newFileContentConstructor.processLine(line)
+	}
+
+	let result = newFileContentConstructor.getResult()
+	return result
+}

evals/diff-edits/openRouterModelsHelper.ts

@@ -0,0 +1,98 @@
+import axios from "axios";
+import path from "path";
+import fs from "fs/promises";
+
+// Minimal type for what we need from OpenRouter model info in evals
+export interface EvalOpenRouterModelInfo {
+    id: string;
+    contextWindow: number;
+    inputPrice?: number; // Price per million tokens
+    outputPrice?: number; // Price per million tokens
+    // Add any other fields if they become necessary for evals
+}
+
+function logHelper(isVerbose: boolean, message: string) {
+    if (isVerbose) {
+        console.log(`[OpenRouterModelsHelper] ${message}`);
+    }
+}
+
+/**
+ * Ensures the cache directory exists within evals and returns its path
+ */
+async function ensureEvalCacheDirectoryExists(): Promise<string> {
+    // Cache directory within evals, e.g., evals/.cache/
+    const cacheDir = path.join(__dirname, "..", ".cache");
+    await fs.mkdir(cacheDir, { recursive: true });
+    return cacheDir;
+}
+
+/**
+ * Fetches, parses, and caches OpenRouter model data.
+ * Tries to read from a local cache first.
+ * @param isVerbose Enable verbose logging
+ * @returns A record of model IDs to their info.
+ */
+export async function loadOpenRouterModelData(isVerbose: boolean = false): Promise<Record<string, EvalOpenRouterModelInfo>> {
+    const cacheDir = await ensureEvalCacheDirectoryExists();
+    const cacheFilePath = path.join(cacheDir, "openRouterModels.json");
+    let models: Record<string, EvalOpenRouterModelInfo> = {};
+
+    try {
+        const stats = await fs.stat(cacheFilePath).catch(() => null);
+        // Use cache if less than 24 hours old
+        if (stats && (Date.now() - stats.mtimeMs < 24 * 60 * 60 * 1000)) {
+            logHelper(isVerbose, "Using cached OpenRouter model data.");
+            const fileContents = await fs.readFile(cacheFilePath, "utf8");
+            models = JSON.parse(fileContents);
+            if (Object.keys(models).length > 0) {
+                return models;
+            }
+            logHelper(isVerbose, "Cache was empty or invalid, fetching fresh data.");
+        } else if (stats) {
+            logHelper(isVerbose, "Cached OpenRouter model data is stale, fetching fresh data.");
+        } else {
+            logHelper(isVerbose, "No cached OpenRouter model data found, fetching fresh data.");
+        }
+    } catch (e) {
+        logHelper(isVerbose, `Error accessing cache, fetching fresh data: ${e}`);
+    }
+
+    try {
+        const response = await axios.get("https://openrouter.ai/api/v1/models");
+        if (response.data?.data) {
+            const rawModels = response.data.data;
+            const parsedModels: Record<string, EvalOpenRouterModelInfo> = {};
+            const parsePrice = (price: any) => price ? parseFloat(price) * 1_000_000 : undefined;
+
+            for (const rawModel of rawModels) {
+                parsedModels[rawModel.id] = {
+                    id: rawModel.id,
+                    contextWindow: rawModel.context_length ?? 0,
+                    inputPrice: parsePrice(rawModel.pricing?.prompt),
+                    outputPrice: parsePrice(rawModel.pricing?.completion),
+                };
+            }
+            await fs.writeFile(cacheFilePath, JSON.stringify(parsedModels, null, 2));
+            logHelper(isVerbose, `Fetched and cached ${Object.keys(parsedModels).length} OpenRouter models.`);
+            return parsedModels;
+        } else {
+            logHelper(isVerbose, "Invalid response structure from OpenRouter API.");
+        }
+    } catch (error) {
+        logHelper(isVerbose, `Error fetching OpenRouter models: ${error}. Attempting to use stale cache if available.`);
+        // Attempt to read stale cache as a last resort if fetching failed
+        try {
+            const fileContents = await fs.readFile(cacheFilePath, "utf8");
+            models = JSON.parse(fileContents);
+            if (Object.keys(models).length > 0) {
+                logHelper(isVerbose, "Successfully loaded stale cache after fetch failure.");
+                return models;
+            }
+        } catch (cacheError) {
+            logHelper(isVerbose, `Failed to read stale cache: ${cacheError}. Proceeding without OpenRouter model data.`);
+        }
+    }
+    // Return empty if all attempts fail, so the caller can decide how to handle it
+    return {};
+}

evals/diff-edits/run_and_open_dashboard.sh

@@ -0,0 +1,34 @@
+#!/bin/bash
+
+# Get the directory of this script to make paths robust
+SCRIPT_DIR=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )
+# The 'evals' directory is the parent of the script's directory
+EVALS_DIR=$(dirname "$SCRIPT_DIR")
+
+# Navigate to the evals directory to ensure npm commands run correctly
+cd "$EVALS_DIR"
+
+# Re-install dependencies and build the CLI
+echo "Ensuring dependencies are up to date and building CLI..."
+npm install && npm run build:cli
+
+# Check if the build was successful before proceeding
+if [ $? -ne 0 ]; then
+    echo "CLI build failed. Aborting evaluation."
+    exit 1
+fi
+
+# Run the evaluation script, passing all arguments from the command line
+echo "Running evaluation..."
+node ./cli/dist/index.js run-diff-eval "$@"
+
+# Check the exit code of the evaluation script
+if [ $? -eq 0 ]; then
+  # If the script succeeded, open the dashboard in the background
+  echo "Evaluation complete. Starting dashboard..."
+  (cd "$SCRIPT_DIR/dashboard" && streamlit run app.py &)
+else
+  # If the script failed, print an error message and exit
+  echo "Evaluation failed. Dashboard will not be started."
+  exit 1
+fi

evals/diff-edits/types.ts

@@ -33,6 +33,7 @@ export interface TestConfig {
 	diff_edit_function: string
 	thinking_tokens_budget: number
 	replay: boolean
+	diff_apply_file?: string
 }
 
 export interface SystemPromptDetails {
@@ -61,7 +62,22 @@ export type ConstructSystemPromptFn = (
 
 export interface TestResult {
 	success: boolean
-	streamResult?: any
+	streamResult?: {
+		assistantMessage: string
+		reasoningMessage: string
+		usage: {
+			inputTokens: number
+			outputTokens: number
+			cacheWriteTokens: number
+			cacheReadTokens: number
+			totalCost: number
+		}
+		timing?: {
+			timeToFirstTokenMs: number
+			timeToFirstEditMs?: number
+			totalRoundTripMs: number
+		}
+	}
 	diffEdit?: string
 	toolCalls?: ExtractedToolCall[]
 	diffEditSuccess?: boolean
@@ -85,4 +101,5 @@ export interface TestInput {
 	diffEditFunction: string
 	thinkingBudgetTokens: number
 	originalDiffEditToolCallMessage?: string
+	diffApplyFile?: string
 }

evals/diff_editing/TestRunner.ts

@@ -1,391 +0,0 @@
-import { runSingleEvaluation, TestInput, TestResult } from "./ClineWrapper"
-import { basicSystemPrompt } from "./prompts/basicSystemPrompt-06-06-25"
-import { claude4SystemPrompt } from "./prompts/claude4SystemPrompt-06-06-25"
-import { formatResponse } from "./helpers"
-import { Anthropic } from "@anthropic-ai/sdk"
-import * as fs from "fs"
-import * as path from "path"
-import { Command } from "commander"
-import { InputMessage, ProcessedTestCase, TestCase, TestConfig, SystemPromptDetails, ConstructSystemPromptFn } from "./types"
-
-function log(isVerbose: boolean, message: string) {
-	if (isVerbose) {
-		console.log(message)
-	}
-}
-
-const systemPromptGeneratorLookup: Record<string, ConstructSystemPromptFn> = {
-	basicSystemPrompt: basicSystemPrompt,
-	claude4SystemPrompt: claude4SystemPrompt,
-}
-
-type TestResultSet = { [test_id: string]: (TestResult & { test_id?: string })[] }
-
-class NodeTestRunner {
-	private apiKey: string | undefined
-
-	constructor(isReplay: boolean) {
-		if (!isReplay) {
-			this.apiKey = process.env.OPENROUTER_API_KEY
-			if (!this.apiKey) {
-				throw new Error("OPENROUTER_API_KEY environment variable not set for a non-replay run.")
-			}
-		}
-	}
-
-	/**
-	 * convert our messages array into a properly formatted Anthropic messages array
-	 */
-	transformMessages(messages: InputMessage[]): Anthropic.Messages.MessageParam[] {
-		return messages.map((msg) => {
-			// Use TextBlockParam here for constructing the input message
-			const content: (Anthropic.TextBlockParam | Anthropic.ImageBlockParam)[] = []
-
-			if (msg.text) {
-				// This object now correctly matches the TextBlockParam type
-				content.push({ type: "text", text: msg.text })
-			}
-
-			if (msg.images && Array.isArray(msg.images)) {
-				const imageBlocks = formatResponse.imageBlocks(msg.images)
-				content.push(...imageBlocks)
-			}
-
-			return {
-				role: msg.role,
-				content: content,
-			}
-		})
-	}
-
-	/**
-	 * Generate the system prompt on the fly
-	 */
-	constructSystemPrompt(systemPromptDetails: SystemPromptDetails, systemPromptName: string) {
-		const systemPromptGenerator = systemPromptGeneratorLookup[systemPromptName]
-
-		const { cwd_value, browser_use, width, height, os_value, shell_value, home_value, mcp_string, user_custom_instructions } =
-			systemPromptDetails
-
-		const systemPrompt = systemPromptGenerator(
-			cwd_value,
-			browser_use,
-			width,
-			height,
-			os_value,
-			shell_value,
-			home_value,
-			mcp_string,
-			user_custom_instructions,
-		)
-
-		return systemPrompt
-	}
-
-	/**
-	 * Loads our test cases from a directory of json files
-	 */
-	loadTestCases(testDirectoryPath: string): TestCase[] {
-		const testCasesArray: TestCase[] = []
-		const dirents = fs.readdirSync(testDirectoryPath, { withFileTypes: true })
-
-		for (const dirent of dirents) {
-			if (dirent.isFile() && dirent.name.endsWith(".json")) {
-				const testFilePath = path.join(testDirectoryPath, dirent.name)
-				const fileContent = fs.readFileSync(testFilePath, "utf8")
-				const testCase: TestCase = JSON.parse(fileContent)
-
-				// Use the filename (without extension) as the test_id if not provided
-				if (!testCase.test_id) {
-					testCase.test_id = path.parse(dirent.name).name
-				}
-				testCasesArray.push(testCase)
-			}
-		}
-
-		return testCasesArray
-	}
-
-	/**
-	 * Saves the test results to the specified output directory.
-	 */
-	saveTestResults(results: TestResultSet, outputPath: string) {
-		// Ensure output directory exists
-		if (!fs.existsSync(outputPath)) {
-			fs.mkdirSync(outputPath, { recursive: true })
-		}
-
-		// Write each test result to its own file
-		for (const testId in results) {
-			const outputFilePath = path.join(outputPath, `${testId}.json`)
-			const testResult = results[testId]
-			fs.writeFileSync(outputFilePath, JSON.stringify(testResult, null, 2))
-		}
-	}
-
-	/**
-	 * Run a single test example
-	 */
-	async runSingleTest(testCase: ProcessedTestCase, testConfig: TestConfig): Promise<TestResult> {
-		if (testConfig.replay && !testCase.original_diff_edit_tool_call_message) {
-			return {
-				success: false,
-				error: "missing_original_diff_edit_tool_call_message",
-				errorString: `Test case ${testCase.test_id} is missing 'original_diff_edit_tool_call_message' for replay.`,
-			}
-		}
-
-		const customSystemPrompt = this.constructSystemPrompt(testCase.system_prompt_details, testConfig.system_prompt_name)
-
-		// messages don't include system prompt and are everything up to the first replace_in_file tool call which results in a diff edit error
-		const input: TestInput = {
-			apiKey: this.apiKey,
-			systemPrompt: customSystemPrompt,
-			messages: testCase.messages,
-			modelId: testConfig.model_id,
-			originalFile: testCase.file_contents,
-			originalFilePath: testCase.file_path,
-			parsingFunction: testConfig.parsing_function,
-			diffEditFunction: testConfig.diff_edit_function,
-			thinkingBudgetTokens: testConfig.thinking_tokens_budget,
-			originalDiffEditToolCallMessage: testConfig.replay ? testCase.original_diff_edit_tool_call_message : undefined,
-		}
-
-		return await runSingleEvaluation(input)
-	}
-
-	/**
-	 * Runs all the text examples synchonously
-	 */
-	async runAllTests(testCases: ProcessedTestCase[], testConfig: TestConfig, isVerbose: boolean): Promise<TestResultSet> {
-		const results: TestResultSet = {}
-
-		for (const testCase of testCases) {
-			results[testCase.test_id] = []
-
-			log(isVerbose, `-Running test: ${testCase.test_id}`)
-			for (let i = 0; i < testConfig.number_of_runs; i++) {
-				const result = await this.runSingleTest(testCase, testConfig)
-				results[testCase.test_id].push(result)
-			}
-		}
-		return results
-	}
-
-	/**
-	 * Runs all of the text examples asynchronously, with concurrency limit
-	 */
-	async runAllTestsParallel(
-		testCases: ProcessedTestCase[],
-		testConfig: TestConfig,
-		isVerbose: boolean,
-		maxConcurrency: number = 20,
-	): Promise<TestResultSet> {
-		const results: TestResultSet = {}
-		testCases.forEach((tc) => {
-			results[tc.test_id] = []
-		})
-
-		// Create a flat list of all individual runs we need to execute
-		const allRuns = testCases.flatMap((testCase) =>
-			Array(testConfig.number_of_runs)
-				.fill(null)
-				.map(() => testCase),
-		)
-
-		for (let i = 0; i < allRuns.length; i += maxConcurrency) {
-			const batch = allRuns.slice(i, i + maxConcurrency)
-
-			const batchPromises = batch.map((testCase) =>
-				this.runSingleTest(testCase, testConfig).then((result) => ({
-					...result,
-					test_id: testCase.test_id,
-				})),
-			)
-
-			const batchResults = await Promise.all(batchPromises)
-
-			// Calculate the total cost for this batch
-			const batchCost = batchResults.reduce((total, result) => {
-				return total + (result.streamResult?.usage?.totalCost || 0)
-			}, 0)
-
-			// Populate the results dictionary
-			for (const result of batchResults) {
-				if (result.test_id) {
-					results[result.test_id].push(result)
-				}
-			}
-
-			const batchNumber = i / maxConcurrency + 1
-			const totalBatches = Math.ceil(allRuns.length / maxConcurrency)
-			log(isVerbose, `-Completed batch ${batchNumber} of ${totalBatches}... (Batch Cost: $${batchCost.toFixed(6)})`)
-		}
-
-		return results
-	}
-
-	/**
-	 * Print output of the tests
-	 */
-	printSummary(results: TestResultSet, isVerbose: boolean) {
-		let totalRuns = 0
-		let totalPasses = 0
-		let totalInputTokens = 0
-		let totalOutputTokens = 0
-		let totalCost = 0
-		let runsWithUsageData = 0
-		let totalDiffEditSuccesses = 0
-		let totalRunsWithToolCalls = 0
-		const testCaseIds = Object.keys(results)
-
-		log(isVerbose, "\n=== TEST SUMMARY ===")
-
-		for (const testId of testCaseIds) {
-			const testResults = results[testId]
-			const passedCount = testResults.filter((r) => r.success && r.diffEditSuccess).length
-			const runCount = testResults.length
-
-			totalRuns += runCount
-			totalPasses += passedCount
-
-			const runsWithToolCalls = testResults.filter((r) => r.success === true).length
-			const diffEditSuccesses = passedCount
-			totalRunsWithToolCalls += runsWithToolCalls
-			totalDiffEditSuccesses += diffEditSuccesses
-
-			// Accumulate token and cost data
-			for (const result of testResults) {
-				if (result.streamResult?.usage) {
-					totalInputTokens += result.streamResult.usage.inputTokens
-					totalOutputTokens += result.streamResult.usage.outputTokens
-					totalCost += result.streamResult.usage.totalCost
-					runsWithUsageData++
-				}
-			}
-
-			log(isVerbose, `\n--- Test Case: ${testId} ---`)
-			log(isVerbose, `  Runs: ${runCount}`)
-			log(isVerbose, `  Passed: ${passedCount}`)
-			log(isVerbose, `  Success Rate: ${runCount > 0 ? ((passedCount / runCount) * 100).toFixed(1) : "N/A"}%`)
-		}
-
-		log(isVerbose, "\n\n=== OVERALL SUMMARY ===")
-		log(isVerbose, `Total Test Cases: ${testCaseIds.length}`)
-		log(isVerbose, `Total Runs Executed: ${totalRuns}`)
-		log(isVerbose, `Overall Passed: ${totalPasses}`)
-		log(isVerbose, `Overall Failed: ${totalRuns - totalPasses}`)
-		log(isVerbose, `Overall Success Rate: ${totalRuns > 0 ? ((totalPasses / totalRuns) * 100).toFixed(1) : "N/A"}%`)
-
-		log(isVerbose, "\n\n=== OVERALL DIFF EDIT SUCCESS RATE ===")
-		if (totalRunsWithToolCalls > 0) {
-			const diffSuccessRate = (totalDiffEditSuccesses / totalRunsWithToolCalls) * 100
-			log(isVerbose, `Total Runs with Successful Tool Calls: ${totalRunsWithToolCalls}`)
-			log(isVerbose, `Total Runs with Successful Diff Edits: ${totalDiffEditSuccesses}`)
-			log(isVerbose, `Diff Edit Success Rate: ${diffSuccessRate.toFixed(1)}%`)
-		} else {
-			log(isVerbose, "No successful tool calls to analyze for diff edit success.")
-		}
-
-		log(isVerbose, "\n\n=== TOKEN & COST ANALYSIS ===")
-		if (runsWithUsageData > 0) {
-			log(isVerbose, `Total Input Tokens: ${totalInputTokens.toLocaleString()}`)
-			log(isVerbose, `Total Output Tokens: ${totalOutputTokens.toLocaleString()}`)
-			log(isVerbose, `Total Cost: $${totalCost.toFixed(6)}`)
-			log(isVerbose, "---")
-			log(
-				isVerbose,
-				`Avg Input Tokens / Run: ${(totalInputTokens / runsWithUsageData).toLocaleString(undefined, {
-					maximumFractionDigits: 0,
-				})}`,
-			)
-			log(
-				isVerbose,
-				`Avg Output Tokens / Run: ${(totalOutputTokens / runsWithUsageData).toLocaleString(undefined, {
-					maximumFractionDigits: 0,
-				})}`,
-			)
-			log(isVerbose, `Avg Cost / Run: $${(totalCost / runsWithUsageData).toFixed(6)}`)
-		} else {
-			log(isVerbose, "No usage data available to analyze.")
-		}
-	}
-}
-
-async function main() {
-	const program = new Command()
-
-	const defaultTestPath = path.join(__dirname, "test_cases")
-	const defaultOutputPath = path.join(__dirname, "test_outputs")
-
-	program
-		.name("TestRunner")
-		.description("Run evaluation tests for diff editing")
-		.version("1.0.0")
-		.option("--test-path <path>", "Path to the directory containing test case JSON files", defaultTestPath)
-		.option("--output-path <path>", "Path to the directory to save the test output JSON files", defaultOutputPath)
-		.option("--model-id <model_id>", "The model ID to use for the test")
-		.option("--system-prompt-name <name>", "The name of the system prompt to use", "basicSystemPrompt")
-		.option("-n, --number-of-runs <number>", "Number of times to run each test case", "1")
-		.option("--parsing-function <name>", "The parsing function to use", "parseAssistantMessageV2")
-		.option("--diff-edit-function <name>", "The diff editing function to use", "constructNewFileContentV2")
-		.option("--thinking-budget <tokens>", "Set the thinking tokens budget", "0")
-		.option("--parallel", "Run tests in parallel", false)
-		.option("--replay", "Run evaluation from a pre-recorded LLM output, skipping the API call", false)
-		.option("-v, --verbose", "Enable verbose logging", false)
-
-	program.parse(process.argv)
-
-	const options = program.opts()
-	const isVerbose = options.verbose
-	const testPath = options.testPath
-	const outputPath = options.outputPath
-
-	const testConfig: TestConfig = {
-		model_id: options.modelId,
-		system_prompt_name: options.systemPromptName,
-		number_of_runs: parseInt(options.numberOfRuns, 10),
-		parsing_function: options.parsingFunction,
-		diff_edit_function: options.diffEditFunction,
-		thinking_tokens_budget: parseInt(options.thinkingBudget, 10),
-		replay: options.replay,
-	}
-
-	try {
-		const startTime = Date.now()
-
-		const runner = new NodeTestRunner(testConfig.replay)
-		const testCases = runner.loadTestCases(testPath)
-
-		const processedTestCases: ProcessedTestCase[] = testCases.map((tc) => ({
-			...tc,
-			messages: runner.transformMessages(tc.messages),
-		}))
-
-		log(isVerbose, `-Loaded ${testCases.length} test cases.`)
-		log(isVerbose, `-Executing ${testConfig.number_of_runs} run(s) per test case.`)
-		if (testConfig.replay) {
-			log(isVerbose, `-Running in REPLAY mode. No API calls will be made.`)
-		}
-		log(isVerbose, "Starting tests...\n")
-
-		const results = options.parallel
-			? await runner.runAllTestsParallel(processedTestCases, testConfig, isVerbose)
-			: await runner.runAllTests(processedTestCases, testConfig, isVerbose)
-
-		runner.printSummary(results, isVerbose)
-
-		const endTime = Date.now()
-		const durationSeconds = ((endTime - startTime) / 1000).toFixed(2)
-		log(isVerbose, `\n-Total execution time: ${durationSeconds} seconds`)
-
-		runner.saveTestResults(results, outputPath)
-	} catch (error) {
-		console.error("\nError running tests:", error)
-		process.exit(1)
-	}
-}
-
-if (require.main === module) {
-	main()
-}

evals/package.json

@@ -0,0 +1,44 @@
+{
+  "name": "cline-evals",
+  "version": "0.1.0",
+  "description": "Evaluation scripts and tools for Cline",
+  "main": "cli/dist/index.js",
+  "scripts": {
+    "build:cli": "cd cli && tsc",
+    "start:cli": "cd cli && node dist/index.js",
+    "dev:cli": "cd cli && ts-node src/index.ts",
+    "diff-eval": "./diff-edits/run_and_open_dashboard.sh",
+    "test": "echo \"Error: no test specified\" && exit 1"
+  },
+  "keywords": [
+    "cline",
+    "evaluation",
+    "benchmark",
+    "diff-edits"
+  ],
+  "author": "",
+  "license": "MIT",
+  "dependencies": {
+    "axios": "^1.8.2",
+    "better-sqlite3": "^11.10.0",
+    "chalk": "^4.1.2",
+    "dotenv": "^16.5.0",
+    "commander": "^9.4.1",
+    "execa": "^5.1.1",
+    "node-fetch": "^2.7.0",
+    "ora": "^5.4.1",
+    "sqlite": "^4.1.2",
+    "tiktoken": "^1.0.21",
+    "uuid": "^9.0.0",
+    "yargs": "^17.6.2"
+  },
+  "devDependencies": {
+    "@types/better-sqlite3": "^7.6.3",
+    "@types/node": "^18.11.18",
+    "@types/node-fetch": "^2.6.12",
+    "@types/uuid": "^9.0.0",
+    "@types/yargs": "^17.0.19",
+    "ts-node": "^10.9.1",
+    "typescript": "^4.9.4"
+  }
+}

evals/tsconfig.json

@@ -0,0 +1,6 @@
+{
+	"extends": "../tsconfig.json",
+	"compilerOptions": {
+		"baseUrl": ".."
+	}
+}

package-lock.json

@@ -1,12 +1,12 @@
 {
 	"name": "claude-dev",
-	"version": "3.17.14",
+	"version": "3.17.16",
 	"lockfileVersion": 2,
 	"requires": true,
 	"packages": {
 		"": {
 			"name": "claude-dev",
-			"version": "3.17.14",
+			"version": "3.17.16",
 			"license": "Apache-2.0",
 			"dependencies": {
 				"@anthropic-ai/sdk": "^0.37.0",
@@ -109,6 +109,7 @@
 				"prettier": "^3.3.3",
 				"protoc-gen-ts": "^0.8.7",
 				"proxyquire": "^2.1.3",
+				"rimraf": "^6.0.1",
 				"should": "^13.2.3",
 				"sinon": "^19.0.2",
 				"ts-node": "^10.9.2",
@@ -3701,6 +3702,29 @@
 				}
 			}
 		},
+		"node_modules/@isaacs/balanced-match": {
+			"version": "4.0.1",
+			"resolved": "https://registry.npmjs.org/@isaacs/balanced-match/-/balanced-match-4.0.1.tgz",
+			"integrity": "sha512-yzMTt9lEb8Gv7zRioUilSglI0c0smZ9k5D65677DLWLtWJaXIS3CqcGyUFByYKlnUj6TkjLVs54fBl6+TiGQDQ==",
+			"dev": true,
+			"license": "MIT",
+			"engines": {
+				"node": "20 || >=22"
+			}
+		},
+		"node_modules/@isaacs/brace-expansion": {
+			"version": "5.0.0",
+			"resolved": "https://registry.npmjs.org/@isaacs/brace-expansion/-/brace-expansion-5.0.0.tgz",
+			"integrity": "sha512-ZT55BDLV0yv0RBm2czMiZ+SqCGO7AvmOM3G/w2xhVPH+te0aKgFjmBvGlL1dH+ql2tgGO3MVrbb3jCKyvpgnxA==",
+			"dev": true,
+			"license": "MIT",
+			"dependencies": {
+				"@isaacs/balanced-match": "^4.0.1"
+			},
+			"engines": {
+				"node": "20 || >=22"
+			}
+		},
 		"node_modules/@isaacs/cliui": {
 			"version": "8.0.2",
 			"resolved": "https://registry.npmjs.org/@isaacs/cliui/-/cliui-8.0.2.tgz",
@@ -4025,6 +4049,17 @@
 				"node": ">= 6.0.0"
 			}
 		},
+		"node_modules/@mapbox/node-pre-gyp/node_modules/brace-expansion": {
+			"version": "1.1.12",
+			"resolved": "https://registry.npmjs.org/brace-expansion/-/brace-expansion-1.1.12.tgz",
+			"integrity": "sha512-9T9UjW3r0UW5c1Q7GTwllptXwhvYmEzFhzMfZ9H7FQWt+uZePjZPjBP/W1ZEyZ1twGWom5/56TF4lPcqjnDHcg==",
+			"dev": true,
+			"license": "MIT",
+			"dependencies": {
+				"balanced-match": "^1.0.0",
+				"concat-map": "0.0.1"
+			}
+		},
 		"node_modules/@mapbox/node-pre-gyp/node_modules/detect-libc": {
 			"version": "2.0.4",
 			"resolved": "https://registry.npmjs.org/detect-libc/-/detect-libc-2.0.4.tgz",
@@ -4035,6 +4070,28 @@
 				"node": ">=8"
 			}
 		},
+		"node_modules/@mapbox/node-pre-gyp/node_modules/glob": {
+			"version": "7.2.3",
+			"resolved": "https://registry.npmjs.org/glob/-/glob-7.2.3.tgz",
+			"integrity": "sha512-nFR0zLpU2YCaRxwoCJvL6UvCH2JFyFVIvwTLsIf21AuHlMskA1hhTdk+LlYJtOlYt9v6dvszD2BGRqBL+iQK9Q==",
+			"deprecated": "Glob versions prior to v9 are no longer supported",
+			"dev": true,
+			"license": "ISC",
+			"dependencies": {
+				"fs.realpath": "^1.0.0",
+				"inflight": "^1.0.4",
+				"inherits": "2",
+				"minimatch": "^3.1.1",
+				"once": "^1.3.0",
+				"path-is-absolute": "^1.0.0"
+			},
+			"engines": {
+				"node": "*"
+			},
+			"funding": {
+				"url": "https://github.com/sponsors/isaacs"
+			}
+		},
 		"node_modules/@mapbox/node-pre-gyp/node_modules/https-proxy-agent": {
 			"version": "5.0.1",
 			"resolved": "https://registry.npmjs.org/https-proxy-agent/-/https-proxy-agent-5.0.1.tgz",
@@ -4075,6 +4132,36 @@
 				"semver": "bin/semver.js"
 			}
 		},
+		"node_modules/@mapbox/node-pre-gyp/node_modules/minimatch": {
+			"version": "3.1.2",
+			"resolved": "https://registry.npmjs.org/minimatch/-/minimatch-3.1.2.tgz",
+			"integrity": "sha512-J7p63hRiAjw1NDEww1W7i37+ByIrOWO5XQQAzZ3VOcL0PNybwpfmV/N05zFAzwQ9USyEcX6t3UO+K5aqBQOIHw==",
+			"dev": true,
+			"license": "ISC",
+			"dependencies": {
+				"brace-expansion": "^1.1.7"
+			},
+			"engines": {
+				"node": "*"
+			}
+		},
+		"node_modules/@mapbox/node-pre-gyp/node_modules/rimraf": {
+			"version": "3.0.2",
+			"resolved": "https://registry.npmjs.org/rimraf/-/rimraf-3.0.2.tgz",
+			"integrity": "sha512-JZkJMZkAGFFPP2YqXZXPbMlMBgsxzE8ILs4lMIX/2o0L9UBw9O/Y3o6wFw/i9YLapcUJWwqbi3kdxIPdC62TIA==",
+			"deprecated": "Rimraf versions prior to v4 are no longer supported",
+			"dev": true,
+			"license": "ISC",
+			"dependencies": {
+				"glob": "^7.1.3"
+			},
+			"bin": {
+				"rimraf": "bin.js"
+			},
+			"funding": {
+				"url": "https://github.com/sponsors/isaacs"
+			}
+		},
 		"node_modules/@mdx-js/mdx": {
 			"version": "3.1.0",
 			"resolved": "https://registry.npmjs.org/@mdx-js/mdx/-/mdx-3.1.0.tgz",
@@ -13169,6 +13256,69 @@
 				"node": "^10.12.0 || >=12.0.0"
 			}
 		},
+		"node_modules/flat-cache/node_modules/brace-expansion": {
+			"version": "1.1.12",
+			"resolved": "https://registry.npmjs.org/brace-expansion/-/brace-expansion-1.1.12.tgz",
+			"integrity": "sha512-9T9UjW3r0UW5c1Q7GTwllptXwhvYmEzFhzMfZ9H7FQWt+uZePjZPjBP/W1ZEyZ1twGWom5/56TF4lPcqjnDHcg==",
+			"dev": true,
+			"license": "MIT",
+			"dependencies": {
+				"balanced-match": "^1.0.0",
+				"concat-map": "0.0.1"
+			}
+		},
+		"node_modules/flat-cache/node_modules/glob": {
+			"version": "7.2.3",
+			"resolved": "https://registry.npmjs.org/glob/-/glob-7.2.3.tgz",
+			"integrity": "sha512-nFR0zLpU2YCaRxwoCJvL6UvCH2JFyFVIvwTLsIf21AuHlMskA1hhTdk+LlYJtOlYt9v6dvszD2BGRqBL+iQK9Q==",
+			"deprecated": "Glob versions prior to v9 are no longer supported",
+			"dev": true,
+			"license": "ISC",
+			"dependencies": {
+				"fs.realpath": "^1.0.0",
+				"inflight": "^1.0.4",
+				"inherits": "2",
+				"minimatch": "^3.1.1",
+				"once": "^1.3.0",
+				"path-is-absolute": "^1.0.0"
+			},
+			"engines": {
+				"node": "*"
+			},
+			"funding": {
+				"url": "https://github.com/sponsors/isaacs"
+			}
+		},
+		"node_modules/flat-cache/node_modules/minimatch": {
+			"version": "3.1.2",
+			"resolved": "https://registry.npmjs.org/minimatch/-/minimatch-3.1.2.tgz",
+			"integrity": "sha512-J7p63hRiAjw1NDEww1W7i37+ByIrOWO5XQQAzZ3VOcL0PNybwpfmV/N05zFAzwQ9USyEcX6t3UO+K5aqBQOIHw==",
+			"dev": true,
+			"license": "ISC",
+			"dependencies": {
+				"brace-expansion": "^1.1.7"
+			},
+			"engines": {
+				"node": "*"
+			}
+		},
+		"node_modules/flat-cache/node_modules/rimraf": {
+			"version": "3.0.2",
+			"resolved": "https://registry.npmjs.org/rimraf/-/rimraf-3.0.2.tgz",
+			"integrity": "sha512-JZkJMZkAGFFPP2YqXZXPbMlMBgsxzE8ILs4lMIX/2o0L9UBw9O/Y3o6wFw/i9YLapcUJWwqbi3kdxIPdC62TIA==",
+			"deprecated": "Rimraf versions prior to v4 are no longer supported",
+			"dev": true,
+			"license": "ISC",
+			"dependencies": {
+				"glob": "^7.1.3"
+			},
+			"bin": {
+				"rimraf": "bin.js"
+			},
+			"funding": {
+				"url": "https://github.com/sponsors/isaacs"
+			}
+		},
 		"node_modules/flatted": {
 			"version": "3.3.1",
 			"resolved": "https://registry.npmjs.org/flatted/-/flatted-3.3.1.tgz",
@@ -13206,12 +13356,12 @@
 			}
 		},
 		"node_modules/foreground-child": {
-			"version": "3.2.1",
-			"resolved": "https://registry.npmjs.org/foreground-child/-/foreground-child-3.2.1.tgz",
-			"integrity": "sha512-PXUUyLqrR2XCWICfv6ukppP96sdFwWbNEnfEMt7jNsISjMsvaLNinAHNDYyvkyU+SZG2BTSbT5NjG+vZslfGTA==",
+			"version": "3.3.1",
+			"resolved": "https://registry.npmjs.org/foreground-child/-/foreground-child-3.3.1.tgz",
+			"integrity": "sha512-gIXjKqtFuWEgzFRJA9WCQeSJLZDjgJUOMCMzxtvFq/37KojM1BFGufqsCy0r4qSQmYLsZYMeyRqzIWOMup03sw==",
 			"license": "ISC",
 			"dependencies": {
-				"cross-spawn": "^7.0.0",
+				"cross-spawn": "^7.0.6",
 				"signal-exit": "^4.0.1"
 			},
 			"engines": {
@@ -20778,66 +20928,106 @@
 			"license": "MIT"
 		},
 		"node_modules/rimraf": {
-			"version": "3.0.2",
-			"resolved": "https://registry.npmjs.org/rimraf/-/rimraf-3.0.2.tgz",
-			"integrity": "sha512-JZkJMZkAGFFPP2YqXZXPbMlMBgsxzE8ILs4lMIX/2o0L9UBw9O/Y3o6wFw/i9YLapcUJWwqbi3kdxIPdC62TIA==",
-			"deprecated": "Rimraf versions prior to v4 are no longer supported",
+			"version": "6.0.1",
+			"resolved": "https://registry.npmjs.org/rimraf/-/rimraf-6.0.1.tgz",
+			"integrity": "sha512-9dkvaxAsk/xNXSJzMgFqqMCuFgt2+KsOFek3TMLfo8NCPfWpBmqwyNn5Y+NX56QUYfCtsyhF3ayiboEoUmJk/A==",
 			"dev": true,
 			"license": "ISC",
 			"dependencies": {
-				"glob": "^7.1.3"
+				"glob": "^11.0.0",
+				"package-json-from-dist": "^1.0.0"
 			},
 			"bin": {
-				"rimraf": "bin.js"
+				"rimraf": "dist/esm/bin.mjs"
+			},
+			"engines": {
+				"node": "20 || >=22"
 			},
 			"funding": {
 				"url": "https://github.com/sponsors/isaacs"
 			}
 		},
-		"node_modules/rimraf/node_modules/brace-expansion": {
-			"version": "1.1.11",
-			"resolved": "https://registry.npmjs.org/brace-expansion/-/brace-expansion-1.1.11.tgz",
-			"integrity": "sha512-iCuPHDFgrHX7H2vEI/5xpz07zSHB00TpugqhmYtVmMO6518mCuRMoOYFldEBl0g187ufozdaHgWKcYFb61qGiA==",
-			"dev": true,
-			"license": "MIT",
-			"dependencies": {
-				"balanced-match": "^1.0.0",
-				"concat-map": "0.0.1"
-			}
-		},
 		"node_modules/rimraf/node_modules/glob": {
-			"version": "7.2.3",
-			"resolved": "https://registry.npmjs.org/glob/-/glob-7.2.3.tgz",
-			"integrity": "sha512-nFR0zLpU2YCaRxwoCJvL6UvCH2JFyFVIvwTLsIf21AuHlMskA1hhTdk+LlYJtOlYt9v6dvszD2BGRqBL+iQK9Q==",
-			"deprecated": "Glob versions prior to v9 are no longer supported",
+			"version": "11.0.3",
+			"resolved": "https://registry.npmjs.org/glob/-/glob-11.0.3.tgz",
+			"integrity": "sha512-2Nim7dha1KVkaiF4q6Dj+ngPPMdfvLJEOpZk/jKiUAkqKebpGAWQXAq9z1xu9HKu5lWfqw/FASuccEjyznjPaA==",
 			"dev": true,
 			"license": "ISC",
 			"dependencies": {
-				"fs.realpath": "^1.0.0",
-				"inflight": "^1.0.4",
-				"inherits": "2",
-				"minimatch": "^3.1.1",
-				"once": "^1.3.0",
-				"path-is-absolute": "^1.0.0"
+				"foreground-child": "^3.3.1",
+				"jackspeak": "^4.1.1",
+				"minimatch": "^10.0.3",
+				"minipass": "^7.1.2",
+				"package-json-from-dist": "^1.0.0",
+				"path-scurry": "^2.0.0"
+			},
+			"bin": {
+				"glob": "dist/esm/bin.mjs"
 			},
 			"engines": {
-				"node": "*"
+				"node": "20 || >=22"
 			},
 			"funding": {
 				"url": "https://github.com/sponsors/isaacs"
 			}
 		},
+		"node_modules/rimraf/node_modules/jackspeak": {
+			"version": "4.1.1",
+			"resolved": "https://registry.npmjs.org/jackspeak/-/jackspeak-4.1.1.tgz",
+			"integrity": "sha512-zptv57P3GpL+O0I7VdMJNBZCu+BPHVQUk55Ft8/QCJjTVxrnJHuVuX/0Bl2A6/+2oyR/ZMEuFKwmzqqZ/U5nPQ==",
+			"dev": true,
+			"license": "BlueOak-1.0.0",
+			"dependencies": {
+				"@isaacs/cliui": "^8.0.2"
+			},
+			"engines": {
+				"node": "20 || >=22"
+			},
+			"funding": {
+				"url": "https://github.com/sponsors/isaacs"
+			}
+		},
+		"node_modules/rimraf/node_modules/lru-cache": {
+			"version": "11.1.0",
+			"resolved": "https://registry.npmjs.org/lru-cache/-/lru-cache-11.1.0.tgz",
+			"integrity": "sha512-QIXZUBJUx+2zHUdQujWejBkcD9+cs94tLn0+YL8UrCh+D5sCXZ4c7LaEH48pNwRY3MLDgqUFyhlCyjJPf1WP0A==",
+			"dev": true,
+			"license": "ISC",
+			"engines": {
+				"node": "20 || >=22"
+			}
+		},
 		"node_modules/rimraf/node_modules/minimatch": {
-			"version": "3.1.2",
-			"resolved": "https://registry.npmjs.org/minimatch/-/minimatch-3.1.2.tgz",
-			"integrity": "sha512-J7p63hRiAjw1NDEww1W7i37+ByIrOWO5XQQAzZ3VOcL0PNybwpfmV/N05zFAzwQ9USyEcX6t3UO+K5aqBQOIHw==",
+			"version": "10.0.3",
+			"resolved": "https://registry.npmjs.org/minimatch/-/minimatch-10.0.3.tgz",
+			"integrity": "sha512-IPZ167aShDZZUMdRk66cyQAW3qr0WzbHkPdMYa8bzZhlHhO3jALbKdxcaak7W9FfT2rZNpQuUu4Od7ILEpXSaw==",
 			"dev": true,
 			"license": "ISC",
 			"dependencies": {
-				"brace-expansion": "^1.1.7"
+				"@isaacs/brace-expansion": "^5.0.0"
 			},
 			"engines": {
-				"node": "*"
+				"node": "20 || >=22"
+			},
+			"funding": {
+				"url": "https://github.com/sponsors/isaacs"
+			}
+		},
+		"node_modules/rimraf/node_modules/path-scurry": {
+			"version": "2.0.0",
+			"resolved": "https://registry.npmjs.org/path-scurry/-/path-scurry-2.0.0.tgz",
+			"integrity": "sha512-ypGJsmGtdXUOeM5u93TyeIEfEhM6s+ljAhrk5vAvSx8uyY/02OvrZnA0YNGUrPXfpJMgI1ODd3nwz8Npx4O4cg==",
+			"dev": true,
+			"license": "BlueOak-1.0.0",
+			"dependencies": {
+				"lru-cache": "^11.0.0",
+				"minipass": "^7.1.2"
+			},
+			"engines": {
+				"node": "20 || >=22"
+			},
+			"funding": {
+				"url": "https://github.com/sponsors/isaacs"
 			}
 		},
 		"node_modules/router": {
@@ -26670,6 +26860,21 @@
 			"dev": true,
 			"requires": {}
 		},
+		"@isaacs/balanced-match": {
+			"version": "4.0.1",
+			"resolved": "https://registry.npmjs.org/@isaacs/balanced-match/-/balanced-match-4.0.1.tgz",
+			"integrity": "sha512-yzMTt9lEb8Gv7zRioUilSglI0c0smZ9k5D65677DLWLtWJaXIS3CqcGyUFByYKlnUj6TkjLVs54fBl6+TiGQDQ==",
+			"dev": true
+		},
+		"@isaacs/brace-expansion": {
+			"version": "5.0.0",
+			"resolved": "https://registry.npmjs.org/@isaacs/brace-expansion/-/brace-expansion-5.0.0.tgz",
+			"integrity": "sha512-ZT55BDLV0yv0RBm2czMiZ+SqCGO7AvmOM3G/w2xhVPH+te0aKgFjmBvGlL1dH+ql2tgGO3MVrbb3jCKyvpgnxA==",
+			"dev": true,
+			"requires": {
+				"@isaacs/balanced-match": "^4.0.1"
+			}
+		},
 		"@isaacs/cliui": {
 			"version": "8.0.2",
 			"resolved": "https://registry.npmjs.org/@isaacs/cliui/-/cliui-8.0.2.tgz",
@@ -26909,12 +27114,36 @@
 						"debug": "4"
 					}
 				},
+				"brace-expansion": {
+					"version": "1.1.12",
+					"resolved": "https://registry.npmjs.org/brace-expansion/-/brace-expansion-1.1.12.tgz",
+					"integrity": "sha512-9T9UjW3r0UW5c1Q7GTwllptXwhvYmEzFhzMfZ9H7FQWt+uZePjZPjBP/W1ZEyZ1twGWom5/56TF4lPcqjnDHcg==",
+					"dev": true,
+					"requires": {
+						"balanced-match": "^1.0.0",
+						"concat-map": "0.0.1"
+					}
+				},
 				"detect-libc": {
 					"version": "2.0.4",
 					"resolved": "https://registry.npmjs.org/detect-libc/-/detect-libc-2.0.4.tgz",
 					"integrity": "sha512-3UDv+G9CsCKO1WKMGw9fwq/SWJYbI0c5Y7LU1AXYoDdbhE2AHQ6N6Nb34sG8Fj7T5APy8qXDCKuuIHd1BR0tVA==",
 					"dev": true
 				},
+				"glob": {
+					"version": "7.2.3",
+					"resolved": "https://registry.npmjs.org/glob/-/glob-7.2.3.tgz",
+					"integrity": "sha512-nFR0zLpU2YCaRxwoCJvL6UvCH2JFyFVIvwTLsIf21AuHlMskA1hhTdk+LlYJtOlYt9v6dvszD2BGRqBL+iQK9Q==",
+					"dev": true,
+					"requires": {
+						"fs.realpath": "^1.0.0",
+						"inflight": "^1.0.4",
+						"inherits": "2",
+						"minimatch": "^3.1.1",
+						"once": "^1.3.0",
+						"path-is-absolute": "^1.0.0"
+					}
+				},
 				"https-proxy-agent": {
 					"version": "5.0.1",
 					"resolved": "https://registry.npmjs.org/https-proxy-agent/-/https-proxy-agent-5.0.1.tgz",
@@ -26941,6 +27170,24 @@
 							"dev": true
 						}
 					}
+				},
+				"minimatch": {
+					"version": "3.1.2",
+					"resolved": "https://registry.npmjs.org/minimatch/-/minimatch-3.1.2.tgz",
+					"integrity": "sha512-J7p63hRiAjw1NDEww1W7i37+ByIrOWO5XQQAzZ3VOcL0PNybwpfmV/N05zFAzwQ9USyEcX6t3UO+K5aqBQOIHw==",
+					"dev": true,
+					"requires": {
+						"brace-expansion": "^1.1.7"
+					}
+				},
+				"rimraf": {
+					"version": "3.0.2",
+					"resolved": "https://registry.npmjs.org/rimraf/-/rimraf-3.0.2.tgz",
+					"integrity": "sha512-JZkJMZkAGFFPP2YqXZXPbMlMBgsxzE8ILs4lMIX/2o0L9UBw9O/Y3o6wFw/i9YLapcUJWwqbi3kdxIPdC62TIA==",
+					"dev": true,
+					"requires": {
+						"glob": "^7.1.3"
+					}
 				}
 			}
 		},
@@ -33440,6 +33687,50 @@
 				"flatted": "^3.2.9",
 				"keyv": "^4.5.3",
 				"rimraf": "^3.0.2"
+			},
+			"dependencies": {
+				"brace-expansion": {
+					"version": "1.1.12",
+					"resolved": "https://registry.npmjs.org/brace-expansion/-/brace-expansion-1.1.12.tgz",
+					"integrity": "sha512-9T9UjW3r0UW5c1Q7GTwllptXwhvYmEzFhzMfZ9H7FQWt+uZePjZPjBP/W1ZEyZ1twGWom5/56TF4lPcqjnDHcg==",
+					"dev": true,
+					"requires": {
+						"balanced-match": "^1.0.0",
+						"concat-map": "0.0.1"
+					}
+				},
+				"glob": {
+					"version": "7.2.3",
+					"resolved": "https://registry.npmjs.org/glob/-/glob-7.2.3.tgz",
+					"integrity": "sha512-nFR0zLpU2YCaRxwoCJvL6UvCH2JFyFVIvwTLsIf21AuHlMskA1hhTdk+LlYJtOlYt9v6dvszD2BGRqBL+iQK9Q==",
+					"dev": true,
+					"requires": {
+						"fs.realpath": "^1.0.0",
+						"inflight": "^1.0.4",
+						"inherits": "2",
+						"minimatch": "^3.1.1",
+						"once": "^1.3.0",
+						"path-is-absolute": "^1.0.0"
+					}
+				},
+				"minimatch": {
+					"version": "3.1.2",
+					"resolved": "https://registry.npmjs.org/minimatch/-/minimatch-3.1.2.tgz",
+					"integrity": "sha512-J7p63hRiAjw1NDEww1W7i37+ByIrOWO5XQQAzZ3VOcL0PNybwpfmV/N05zFAzwQ9USyEcX6t3UO+K5aqBQOIHw==",
+					"dev": true,
+					"requires": {
+						"brace-expansion": "^1.1.7"
+					}
+				},
+				"rimraf": {
+					"version": "3.0.2",
+					"resolved": "https://registry.npmjs.org/rimraf/-/rimraf-3.0.2.tgz",
+					"integrity": "sha512-JZkJMZkAGFFPP2YqXZXPbMlMBgsxzE8ILs4lMIX/2o0L9UBw9O/Y3o6wFw/i9YLapcUJWwqbi3kdxIPdC62TIA==",
+					"dev": true,
+					"requires": {
+						"glob": "^7.1.3"
+					}
+				}
 			}
 		},
 		"flatted": {
@@ -33463,11 +33754,11 @@
 			}
 		},
 		"foreground-child": {
-			"version": "3.2.1",
-			"resolved": "https://registry.npmjs.org/foreground-child/-/foreground-child-3.2.1.tgz",
-			"integrity": "sha512-PXUUyLqrR2XCWICfv6ukppP96sdFwWbNEnfEMt7jNsISjMsvaLNinAHNDYyvkyU+SZG2BTSbT5NjG+vZslfGTA==",
+			"version": "3.3.1",
+			"resolved": "https://registry.npmjs.org/foreground-child/-/foreground-child-3.3.1.tgz",
+			"integrity": "sha512-gIXjKqtFuWEgzFRJA9WCQeSJLZDjgJUOMCMzxtvFq/37KojM1BFGufqsCy0r4qSQmYLsZYMeyRqzIWOMup03sw==",
 			"requires": {
-				"cross-spawn": "^7.0.0",
+				"cross-spawn": "^7.0.6",
 				"signal-exit": "^4.0.1"
 			}
 		},
@@ -38623,45 +38914,61 @@
 			"dev": true
 		},
 		"rimraf": {
-			"version": "3.0.2",
-			"resolved": "https://registry.npmjs.org/rimraf/-/rimraf-3.0.2.tgz",
-			"integrity": "sha512-JZkJMZkAGFFPP2YqXZXPbMlMBgsxzE8ILs4lMIX/2o0L9UBw9O/Y3o6wFw/i9YLapcUJWwqbi3kdxIPdC62TIA==",
+			"version": "6.0.1",
+			"resolved": "https://registry.npmjs.org/rimraf/-/rimraf-6.0.1.tgz",
+			"integrity": "sha512-9dkvaxAsk/xNXSJzMgFqqMCuFgt2+KsOFek3TMLfo8NCPfWpBmqwyNn5Y+NX56QUYfCtsyhF3ayiboEoUmJk/A==",
 			"dev": true,
 			"requires": {
-				"glob": "^7.1.3"
+				"glob": "^11.0.0",
+				"package-json-from-dist": "^1.0.0"
 			},
 			"dependencies": {
-				"brace-expansion": {
-					"version": "1.1.11",
-					"resolved": "https://registry.npmjs.org/brace-expansion/-/brace-expansion-1.1.11.tgz",
-					"integrity": "sha512-iCuPHDFgrHX7H2vEI/5xpz07zSHB00TpugqhmYtVmMO6518mCuRMoOYFldEBl0g187ufozdaHgWKcYFb61qGiA==",
+				"glob": {
+					"version": "11.0.3",
+					"resolved": "https://registry.npmjs.org/glob/-/glob-11.0.3.tgz",
+					"integrity": "sha512-2Nim7dha1KVkaiF4q6Dj+ngPPMdfvLJEOpZk/jKiUAkqKebpGAWQXAq9z1xu9HKu5lWfqw/FASuccEjyznjPaA==",
 					"dev": true,
 					"requires": {
-						"balanced-match": "^1.0.0",
-						"concat-map": "0.0.1"
+						"foreground-child": "^3.3.1",
+						"jackspeak": "^4.1.1",
+						"minimatch": "^10.0.3",
+						"minipass": "^7.1.2",
+						"package-json-from-dist": "^1.0.0",
+						"path-scurry": "^2.0.0"
 					}
 				},
-				"glob": {
-					"version": "7.2.3",
-					"resolved": "https://registry.npmjs.org/glob/-/glob-7.2.3.tgz",
-					"integrity": "sha512-nFR0zLpU2YCaRxwoCJvL6UvCH2JFyFVIvwTLsIf21AuHlMskA1hhTdk+LlYJtOlYt9v6dvszD2BGRqBL+iQK9Q==",
+				"jackspeak": {
+					"version": "4.1.1",
+					"resolved": "https://registry.npmjs.org/jackspeak/-/jackspeak-4.1.1.tgz",
+					"integrity": "sha512-zptv57P3GpL+O0I7VdMJNBZCu+BPHVQUk55Ft8/QCJjTVxrnJHuVuX/0Bl2A6/+2oyR/ZMEuFKwmzqqZ/U5nPQ==",
 					"dev": true,
 					"requires": {
-						"fs.realpath": "^1.0.0",
-						"inflight": "^1.0.4",
-						"inherits": "2",
-						"minimatch": "^3.1.1",
-						"once": "^1.3.0",
-						"path-is-absolute": "^1.0.0"
+						"@isaacs/cliui": "^8.0.2"
 					}
 				},
+				"lru-cache": {
+					"version": "11.1.0",
+					"resolved": "https://registry.npmjs.org/lru-cache/-/lru-cache-11.1.0.tgz",
+					"integrity": "sha512-QIXZUBJUx+2zHUdQujWejBkcD9+cs94tLn0+YL8UrCh+D5sCXZ4c7LaEH48pNwRY3MLDgqUFyhlCyjJPf1WP0A==",
+					"dev": true
+				},
 				"minimatch": {
-					"version": "3.1.2",
-					"resolved": "https://registry.npmjs.org/minimatch/-/minimatch-3.1.2.tgz",
-					"integrity": "sha512-J7p63hRiAjw1NDEww1W7i37+ByIrOWO5XQQAzZ3VOcL0PNybwpfmV/N05zFAzwQ9USyEcX6t3UO+K5aqBQOIHw==",
+					"version": "10.0.3",
+					"resolved": "https://registry.npmjs.org/minimatch/-/minimatch-10.0.3.tgz",
+					"integrity": "sha512-IPZ167aShDZZUMdRk66cyQAW3qr0WzbHkPdMYa8bzZhlHhO3jALbKdxcaak7W9FfT2rZNpQuUu4Od7ILEpXSaw==",
 					"dev": true,
 					"requires": {
-						"brace-expansion": "^1.1.7"
+						"@isaacs/brace-expansion": "^5.0.0"
+					}
+				},
+				"path-scurry": {
+					"version": "2.0.0",
+					"resolved": "https://registry.npmjs.org/path-scurry/-/path-scurry-2.0.0.tgz",
+					"integrity": "sha512-ypGJsmGtdXUOeM5u93TyeIEfEhM6s+ljAhrk5vAvSx8uyY/02OvrZnA0YNGUrPXfpJMgI1ODd3nwz8Npx4O4cg==",
+					"dev": true,
+					"requires": {
+						"lru-cache": "^11.0.0",
+						"minipass": "^7.1.2"
 					}
 				}
 			}

package.json

@@ -2,7 +2,7 @@
 	"name": "claude-dev",
 	"displayName": "Cline",
 	"description": "Autonomous coding agent right in your IDE, capable of creating/editing files, running commands, using the browser, and more with your permission every step of the way.",
-	"version": "3.17.14",
+	"version": "3.17.16",
 	"icon": "assets/icons/icon.png",
 	"engines": {
 		"vscode": "^1.84.0"
@@ -332,6 +332,7 @@
 		"package": "npm run check-types && npm run build:webview && npm run lint && node esbuild.js --production",
 		"protos": "node proto/build-proto.js && node scripts/generate-server-setup.mjs && node scripts/generate-host-bridge-client.mjs",
 		"postprotos": "prettier src/shared/proto src/core/controller src/hosts/ webview-ui/src/services src/generated --write --log-level warn",
+		"clean": "rimraf dist dist-standalone webview-ui/build src/generated",
 		"compile-tests": "node ./scripts/build-tests.js",
 		"watch-tests": "tsc -p . -w --outDir out",
 		"pretest": "npm run compile-tests && npm run compile && npm run compile-standalone && npm run lint",
@@ -396,6 +397,7 @@
 		"prettier": "^3.3.3",
 		"protoc-gen-ts": "^0.8.7",
 		"proxyquire": "^2.1.3",
+		"rimraf": "^6.0.1",
 		"should": "^13.2.3",
 		"sinon": "^19.0.2",
 		"ts-node": "^10.9.2",

proto/account.proto

@@ -19,4 +19,58 @@ service AccountService {
 
     // Subscribe to auth callback events (when authentication tokens are received)
     rpc subscribeToAuthCallback(EmptyRequest) returns (stream String);
+    
+    // Handles authentication state changes from the Firebase context.
+    // Updates the user info in global state and returns the updated value.
+    rpc authStateChanged(AuthStateChangedRequest) returns (AuthStateChanged);
+    
+    // Fetches all user credits data (balance, usage transactions, payment transactions)
+    rpc fetchUserCreditsData(EmptyRequest) returns (UserCreditsData);
+}
+
+message AuthStateChangedRequest {
+    Metadata metadata = 1;
+    UserInfo user = 2;
+}
+
+message AuthStateChanged {
+    optional UserInfo user = 1;
+}
+
+message UserInfo {
+    optional string display_name = 1;
+    optional string email = 2;
+    optional string photo_url = 3;
+}
+
+// Response containing all user credits data
+message UserCreditsData {
+    UserCreditsBalance balance = 1;
+    repeated UsageTransaction usage_transactions = 2;
+    repeated PaymentTransaction payment_transactions = 3;
+}
+
+// User's current credit balance
+message UserCreditsBalance {
+    double current_balance = 1;
+}
+
+// Usage transaction record
+message UsageTransaction {
+    string spent_at = 1;
+    string creator_id = 2;
+    double credits = 3;
+    string model_provider = 4;
+    string model = 5;
+    int32 prompt_tokens = 6;
+    int32 completion_tokens = 7;
+    int32 total_tokens = 8;
+}
+
+// Payment transaction record
+message PaymentTransaction {
+    string paid_at = 1;
+    string creator_id = 2;
+    int32 amount_cents = 3;
+    double credits = 4;
 }

proto/file.proto

@@ -29,9 +29,6 @@ service FileService {
   // Search git commits in the workspace
   rpc searchCommits(StringRequest) returns (GitCommits);
 
-  // Select images from the file system and return as data URLs
-  rpc selectImages(EmptyRequest) returns (StringArray);
-
   // Select images and other files from the file system and returns as data URLs & paths respectively
   rpc selectFiles(BooleanRequest) returns (StringArrays);
   

proto/models.proto

@@ -123,6 +123,7 @@ enum ApiProvider {
   CEREBRAS = 23;
   SAPAICORE = 24;
   CLAUDE_CODE = 25;
+  GEMINI_CLI = 26;
 }
 
 // Model info for OpenAI-compatible models
@@ -236,4 +237,5 @@ message ModelsApiConfiguration {
   optional string sap_ai_core_token_url = 71;
   optional string sap_ai_core_base_url = 72;
   optional string claude_code_path = 73;
-}
+  optional string gemini_cli_path = 74;
+}

proto/state.proto

@@ -17,6 +17,7 @@ service StateService {
   rpc togglePlanActMode(TogglePlanActModeRequest) returns (Boolean);
   rpc updateAutoApprovalSettings(AutoApprovalSettingsRequest) returns (Empty);
   rpc updateSettings(UpdateSettingsRequest) returns (Empty);
+  rpc updateTelemetrySetting(TelemetrySettingRequest) returns (Empty);
 }
 
 message State {
@@ -88,6 +89,17 @@ message AutoApprovalSettingsRequest {
   repeated string favorites = 7;
 }
 
+enum TelemetrySettingEnum {
+  UNSET = 0;
+  ENABLED = 1;
+  DISABLED = 2;
+}
+
+message TelemetrySettingRequest {
+  Metadata metadata = 1;
+  TelemetrySettingEnum setting = 2;
+}
+
 // Message for updating settings
 message UpdateSettingsRequest {
   Metadata metadata = 1;

proto/task.proto

@@ -35,6 +35,8 @@ service TaskService {
   rpc taskCompletionViewChanges(Int64Request) returns (Empty);
   // Executes a quick win task with command and title
   rpc executeQuickWin(ExecuteQuickWinRequest) returns (Empty);
+  // Deletes all task history
+  rpc deleteAllTaskHistory(BooleanRequest) returns (DeleteAllTaskHistoryCount);
 }
 
 // Request message for creating a new task
@@ -116,3 +118,8 @@ message ExecuteQuickWinRequest {
   string command = 2;
   string title = 3;
 }
+
+// Results returned when deleting all task history
+message DeleteAllTaskHistoryCount {
+  int32 tasks_deleted = 1;
+}

proto/ui.proto

@@ -262,4 +262,10 @@ service UiService {
   
   // Subscribe to webview visibility change events
   rpc subscribeToDidBecomeVisible(EmptyRequest) returns (stream Empty);
+
+  // Returns the HTML for the webview index page. This is only used by external clients, not by the vscode webview.
+  rpc getWebviewHtml(EmptyRequest) returns (String);
+  
+  // Opens a URL in the default browser
+  rpc openUrl(StringRequest) returns (Empty);
 }

scripts/get-vscode-usages.sh

@@ -13,7 +13,8 @@ grep -Ev '//.*vscode' | # remove commented out code
 sed 's|.*vscode\.|vscode.|'| # remove everything before vscode.
 sed 's/[^a-zA-Z0-9_.].*$//' | # remove everything after last identifier
 grep -E '\.[a-z][^.]+$' | # remove types (last part of identifier should be lowercase)
-sort | uniq > $SDK_DEST
+sort | uniq -c | sort -n | # Count occurrences
+cat > $SDK_DEST
 }
 echo Wrote uses of the vscode SDK to $(realpath $SDK_DEST)
 

scripts/package-standalone.mjs

@@ -56,13 +56,24 @@ archive.glob("**/*", {
 
 // Add the whole cline directory under "extension"
 archive.directory(process.cwd(), "extension", (entry) => {
-	// Skip certain directories
-	if (
-		entry.name.startsWith(BUILD_DIR + "/") ||
-		entry.name.startsWith("node_modules/") || // node_modules nearly 1GB.
-		entry.name.startsWith("webview-ui/node_modules/") || // node_modules nearly 1GB.
-		entry.name.match(/(^|\/)\./) // exclude dot directories
-	) {
+	// Skip certain directories.
+	const exclude = [
+		BUILD_DIR + "/",
+		"node_modules/", // node_modules nearly 1GB.
+		"webview-ui/node_modules/", // node_modules nearly 1GB.
+	]
+	// These node modules are used at runtime as assets, they need to be included.
+	const include = ["node_modules/@vscode/", "webview-ui/node_modules/katex"]
+	const name = entry.name
+
+	if (include.some((prefix) => name.startsWith(prefix))) {
+		return entry
+	}
+	if (exclude.some((prefix) => name.startsWith(prefix))) {
+		return false
+	}
+	if (name.match(/(^|\/)\./)) {
+		// exclude dot directories
 		return false
 	}
 	return entry

src/api/index.ts

@@ -27,6 +27,7 @@ import { SambanovaHandler } from "./providers/sambanova"
 import { CerebrasHandler } from "./providers/cerebras"
 import { SapAiCoreHandler } from "./providers/sapaicore"
 import { ClaudeCodeHandler } from "./providers/claude-code"
+import { GeminiCliHandler } from "./providers/gemini-cli"
 
 export interface ApiHandler {
 	createMessage(systemPrompt: string, messages: Anthropic.Messages.MessageParam[]): ApiStream
@@ -38,8 +39,7 @@ export interface SingleCompletionHandler {
 	completePrompt(prompt: string): Promise<string>
 }
 
-export function buildApiHandler(configuration: ApiConfiguration): ApiHandler {
-	const { apiProvider, ...options } = configuration
+function createHandlerForProvider(apiProvider: string | undefined, options: any): ApiHandler {
 	switch (apiProvider) {
 		case "anthropic":
 			return new AnthropicHandler(options)
@@ -93,7 +93,33 @@ export function buildApiHandler(configuration: ApiConfiguration): ApiHandler {
 			return new SapAiCoreHandler(options)
 		case "claude-code":
 			return new ClaudeCodeHandler(options)
+		case "gemini-cli":
+			return new GeminiCliHandler(options)
 		default:
 			return new AnthropicHandler(options)
 	}
 }
+
+export function buildApiHandler(configuration: ApiConfiguration): ApiHandler {
+	const { apiProvider, ...options } = configuration
+
+	// Validate thinking budget tokens against model's maxTokens to prevent API errors
+	// wrapped in a try-catch for safety, but this should never throw
+	try {
+		if (options.thinkingBudgetTokens && options.thinkingBudgetTokens > 0) {
+			const handler = createHandlerForProvider(apiProvider, options)
+
+			const modelInfo = handler.getModel().info
+			if (modelInfo.maxTokens && options.thinkingBudgetTokens > modelInfo.maxTokens) {
+				const clippedValue = modelInfo.maxTokens - 1
+				options.thinkingBudgetTokens = clippedValue
+			} else {
+				return handler // don't rebuild unless its necessary
+			}
+		}
+	} catch (error) {
+		console.error("buildApiHandler error:", error)
+	}
+
+	return createHandlerForProvider(apiProvider, options)
+}

src/api/providers/claude-code.ts

@@ -4,7 +4,7 @@ import { type ApiHandler } from ".."
 import { ApiStreamUsageChunk, type ApiStream } from "../transform/stream"
 import { withRetry } from "../retry"
 import { runClaudeCode } from "@/integrations/claude-code/run"
-import { ClaudeCodeMessage } from "@/integrations/claude-code/types"
+import { filterMessagesForClaudeCode } from "@/integrations/claude-code/message-filter"
 
 export class ClaudeCodeHandler implements ApiHandler {
 	private options: ApiHandlerOptions
@@ -19,39 +19,16 @@ export class ClaudeCodeHandler implements ApiHandler {
 		maxDelay: 15000,
 	})
 	async *createMessage(systemPrompt: string, messages: Anthropic.Messages.MessageParam[]): ApiStream {
+		// Filter out image blocks since Claude Code doesn't support them
+		const filteredMessages = filterMessagesForClaudeCode(messages)
+
 		const claudeProcess = runClaudeCode({
 			systemPrompt,
-			messages,
+			messages: filteredMessages,
 			path: this.options.claudeCodePath,
 			modelId: this.getModel().id,
 		})
 
-		const dataQueue: string[] = []
-		let processError = null
-		let errorOutput = ""
-		let exitCode: number | null = null
-
-		claudeProcess.stdout.on("data", (data) => {
-			const output = data.toString()
-			const lines = output.split("\n").filter((line: string) => line.trim() !== "")
-
-			for (const line of lines) {
-				dataQueue.push(line)
-			}
-		})
-
-		claudeProcess.stderr.on("data", (data) => {
-			errorOutput += data.toString()
-		})
-
-		claudeProcess.on("close", (code) => {
-			exitCode = code
-		})
-
-		claudeProcess.on("error", (error) => {
-			processError = error
-		})
-
 		// Usage is included with assistant messages,
 		// but cost is included in the result chunk
 		let usage: ApiStreamUsageChunk = {
@@ -62,61 +39,75 @@ export class ClaudeCodeHandler implements ApiHandler {
 			cacheWriteTokens: 0,
 		}
 
-		while (exitCode !== 0 || dataQueue.length > 0) {
-			if (dataQueue.length === 0) {
-				await new Promise((resolve) => setImmediate(resolve))
-			}
+		let isPaidUsage = true
 
-			if (exitCode !== null && exitCode !== 0) {
-				throw new Error(
-					`Claude Code process exited with code ${exitCode}.${errorOutput ? ` Error output: ${errorOutput.trim()}` : ""}`,
-				)
-			}
-
-			const data = dataQueue.shift()
-			if (!data) {
-				continue
-			}
-
-			const chunk = this.attemptParseChunk(data)
-
-			if (!chunk) {
+		for await (const chunk of claudeProcess) {
+			if (typeof chunk === "string") {
 				yield {
 					type: "text",
-					text: data || "",
+					text: chunk,
 				}
 
 				continue
 			}
 
 			if (chunk.type === "system" && chunk.subtype === "init") {
+				// Based on my tests, subscription usage sets the `apiKeySource` to "none"
+				isPaidUsage = chunk.apiKeySource !== "none"
 				continue
 			}
 
 			if (chunk.type === "assistant" && "message" in chunk) {
 				const message = chunk.message
 
-				if (message.stop_reason !== null && message.stop_reason !== "tool_use") {
-					const errorMessage = message.content[0]?.text || `Claude Code stopped with reason: ${message.stop_reason}`
+				if (message.stop_reason !== null) {
+					const content = "text" in message.content[0] ? message.content[0] : undefined
 
-					if (errorMessage.includes("Invalid model name")) {
-						throw new Error(
-							errorMessage +
-								`\n\nAPI keys and subscription plans allow different models. Make sure the selected model is included in your plan.`,
-						)
+					const isError = content && content.text.startsWith(`API Error`)
+					if (isError) {
+						// Error messages are formatted as: `API Error: <<status code>> <<json>>`
+						const errorMessageStart = content.text.indexOf("{")
+						const errorMessage = content.text.slice(errorMessageStart)
+
+						const error = this.attemptParse(errorMessage)
+						if (!error) {
+							throw new Error(content.text)
+						}
+
+						if (error.error.message.includes("Invalid model name")) {
+							throw new Error(
+								content.text +
+									`\n\nAPI keys and subscription plans allow different models. Make sure the selected model is included in your plan.`,
+							)
+						}
+
+						throw new Error(errorMessage)
 					}
-
-					throw new Error(errorMessage)
 				}
 
 				for (const content of message.content) {
-					if (content.type === "text") {
-						yield {
-							type: "text",
-							text: content.text,
-						}
-					} else {
-						console.warn("Unsupported content type:", content.type)
+					switch (content.type) {
+						case "text":
+							yield {
+								type: "text",
+								text: content.text,
+							}
+							break
+						case "thinking":
+							yield {
+								type: "reasoning",
+								reasoning: content.thinking || "",
+							}
+							break
+						case "redacted_thinking":
+							yield {
+								type: "reasoning",
+								reasoning: "[Redacted thinking block]",
+							}
+							break
+						case "tool_use":
+							console.error(`tool_use is not supported yet. Received: ${JSON.stringify(content)}`)
+							break
 					}
 				}
 
@@ -129,14 +120,18 @@ export class ClaudeCodeHandler implements ApiHandler {
 			}
 
 			if (chunk.type === "result" && "result" in chunk) {
-				usage.totalCost = chunk.cost_usd || 0
+				usage.totalCost = isPaidUsage ? chunk.total_cost_usd : 0
 
 				yield usage
 			}
+		}
+	}
 
-			if (processError) {
-				throw processError
-			}
+	private attemptParse(str: string) {
+		try {
+			return JSON.parse(str)
+		} catch (err) {
+			return null
 		}
 	}
 
@@ -152,14 +147,4 @@ export class ClaudeCodeHandler implements ApiHandler {
 			info: claudeCodeModels[claudeCodeDefaultModelId],
 		}
 	}
-
-	// TOOD: Validate instead of parsing
-	private attemptParseChunk(data: string): ClaudeCodeMessage | null {
-		try {
-			return JSON.parse(data)
-		} catch (error) {
-			console.error("Error parsing chunk:", error)
-			return null
-		}
-	}
 }

src/api/providers/gemini-cli.ts

@@ -0,0 +1,83 @@
+import type { Anthropic } from "@anthropic-ai/sdk"
+import { geminiDefaultModelId, GeminiModelId, geminiModels, type ApiHandlerOptions } from "@/shared/api"
+import { type ApiHandler } from ".."
+import { type ApiStream } from "../transform/stream"
+import { withRetry } from "../retry"
+import { runGeminiCli } from "@/integrations/gemini-cli/run"
+import { convertAnthropicToGeminiCliFormat } from "@/integrations/gemini-cli/message-converter"
+
+export class GeminiCliHandler implements ApiHandler {
+	private options: ApiHandlerOptions
+
+	constructor(options: ApiHandlerOptions) {
+		this.options = options
+	}
+
+	@withRetry({
+		maxRetries: 4,
+		baseDelay: 2000,
+		maxDelay: 15000,
+	})
+	async *createMessage(systemPrompt: string, messages: Anthropic.Messages.MessageParam[]): ApiStream {
+		// Convert messages to Gemini CLI format
+		const prompt = convertAnthropicToGeminiCliFormat(systemPrompt, messages)
+
+		console.log("[GeminiCliHandler] Starting Gemini CLI with prompt length:", prompt.length)
+		console.log("[GeminiCliHandler] Model:", this.getModel().id)
+		console.log("[GeminiCliHandler] Path:", this.options.geminiCliPath || "gemini")
+
+		const geminiProcess = runGeminiCli({
+			prompt,
+			path: this.options.geminiCliPath,
+			modelId: this.getModel().id,
+		})
+
+		let totalText = ""
+		let hasYieldedContent = false
+
+		for await (const chunk of geminiProcess) {
+			if (typeof chunk === "string") {
+				totalText += chunk
+				hasYieldedContent = true
+				console.log("[GeminiCliHandler] Received chunk:", chunk.substring(0, 100) + "...")
+				yield {
+					type: "text",
+					text: chunk,
+				}
+			} else if (chunk.type === "error") {
+				console.error("[GeminiCliHandler] Error from Gemini CLI:", chunk.message)
+				throw new Error(chunk.message)
+			}
+		}
+
+		// If no content was yielded, something went wrong
+		if (!hasYieldedContent) {
+			throw new Error("Gemini CLI did not return any content")
+		}
+
+		// Since Gemini CLI doesn't provide token usage in non-interactive mode,
+		// we'll estimate based on the text length
+		const estimatedTokens = Math.ceil(totalText.length / 4)
+		yield {
+			type: "usage",
+			inputTokens: Math.ceil(prompt.length / 4),
+			outputTokens: estimatedTokens,
+			cacheReadTokens: 0,
+			cacheWriteTokens: 0,
+			totalCost: undefined, // Cost calculation would require actual token counts
+		}
+	}
+
+	getModel() {
+		const modelId = this.options.apiModelId
+		if (modelId && modelId in geminiModels) {
+			const id = modelId as GeminiModelId
+			return { id, info: geminiModels[id] }
+		}
+
+		return {
+			id: geminiDefaultModelId,
+			info: geminiModels[geminiDefaultModelId],
+		}
+	}
+}

src/api/providers/litellm.ts

@@ -101,14 +101,21 @@ export class LiteLlmHandler implements ApiHandler {
 			return message
 		})
 
-		const stream = await this.client.chat.completions.create({
+		const requestPayload: OpenAI.Chat.Completions.ChatCompletionCreateParamsStreaming & {
+			metadata?: { cline_task_id: string }
+		} = {
 			model: this.options.liteLlmModelId || liteLlmDefaultModelId,
 			messages: [enhancedSystemMessage, ...enhancedMessages],
 			temperature,
 			stream: true,
 			stream_options: { include_usage: true },
 			...(thinkingConfig && { thinking: thinkingConfig }), // Add thinking configuration when applicable
-		})
+			...(this.options.taskId && {
+				metadata: { cline_task_id: this.options.taskId },
+			}),
+		}
+
+		const stream = await this.client.chat.completions.create(requestPayload)
 
 		const inputCost = (await this.calculateCost(1e6, 0)) || 0
 		const outputCost = (await this.calculateCost(0, 1e6)) || 0

src/core/controller/account/authStateChanged.ts

@@ -0,0 +1,23 @@
+import { AuthStateChangedRequest, AuthStateChanged } from "@shared/proto/account"
+import type { Controller } from "../index"
+import { updateGlobalState } from "../../storage/state"
+
+/**
+ * Handles authentication state changes from the Firebase context.
+ * Updates the user info in global state and returns the updated value.
+ * @param controller The controller instance
+ * @param request The auth state change request
+ * @returns The updated user info
+ */
+export async function authStateChanged(controller: Controller, request: AuthStateChangedRequest): Promise<AuthStateChanged> {
+	try {
+		// Store the user info directly in global state
+		await updateGlobalState(controller.context, "userInfo", request.user)
+
+		// Return the same user info
+		return AuthStateChanged.create({ user: request.user })
+	} catch (error) {
+		console.error(`Failed to update auth state: ${error}`)
+		throw error
+	}
+}

src/core/controller/account/fetchUserCreditsData.ts

@@ -0,0 +1,34 @@
+import type { Controller } from "../index"
+import type { EmptyRequest } from "@shared/proto/common"
+import { UserCreditsData } from "@shared/proto/account"
+
+/**
+ * Handles fetching all user credits data (balance, usage, payments)
+ * @param controller The controller instance
+ * @param request Empty request
+ * @returns User credits data response
+ */
+export async function fetchUserCreditsData(controller: Controller, request: EmptyRequest): Promise<UserCreditsData> {
+	try {
+		if (!controller.accountService) {
+			throw new Error("Account service not available")
+		}
+
+		// Call the individual RPC variants in parallel
+		const [balance, usageTransactions, paymentTransactions] = await Promise.all([
+			controller.accountService.fetchBalanceRPC(),
+			controller.accountService.fetchUsageTransactionsRPC(),
+			controller.accountService.fetchPaymentTransactionsRPC(),
+		])
+
+		// Since generated types match exactly, no conversion needed!
+		return UserCreditsData.create({
+			balance: balance ? { currentBalance: balance.currentBalance } : { currentBalance: 0 },
+			usageTransactions: usageTransactions || [],
+			paymentTransactions: paymentTransactions || [],
+		})
+	} catch (error) {
+		console.error(`Failed to fetch user credits data: ${error}`)
+		throw error
+	}
+}

src/core/controller/file/selectImages.ts

@@ -1,21 +0,0 @@
-import { Controller } from ".."
-import { EmptyRequest, StringArray } from "@shared/proto/common"
-import { selectImages as selectImagesIntegration } from "@integrations/misc/process-images"
-import { FileMethodHandler } from "./index"
-
-/**
- * Prompts the user to select images from the file system and returns them as data URLs
- * @param controller The controller instance
- * @param request Empty request, no parameters needed
- * @returns Array of image data URLs
- */
-export const selectImages: FileMethodHandler = async (controller: Controller, request: EmptyRequest): Promise<StringArray> => {
-	try {
-		const images = await selectImagesIntegration()
-		return StringArray.create({ values: images })
-	} catch (error) {
-		console.error("Error selecting images:", error)
-		// Return empty array on error
-		return StringArray.create({ values: [] })
-	}
-}

src/core/controller/index.ts

@@ -16,10 +16,11 @@ import { McpHub } from "@services/mcp/McpHub"
 import { telemetryService } from "@/services/posthog/telemetry/TelemetryService"
 import { ApiProvider, ModelInfo } from "@shared/api"
 import { ChatContent } from "@shared/ChatContent"
-import { ChatSettings } from "@shared/ChatSettings"
+import { ChatSettings, StoredChatSettings } from "@shared/ChatSettings"
 import { ExtensionMessage, ExtensionState, Platform } from "@shared/ExtensionMessage"
 import { HistoryItem } from "@shared/HistoryItem"
 import { McpMarketplaceCatalog } from "@shared/mcp"
+import { UserInfo } from "@shared/UserInfo"
 import { TelemetrySetting } from "@shared/TelemetrySetting"
 import { WebviewMessage } from "@shared/WebviewMessage"
 import { fileExistsAtPath } from "@utils/fs"
@@ -43,6 +44,8 @@ import { sendAddToInputEvent } from "./ui/subscribeToAddToInput"
 import { sendAuthCallbackEvent } from "./account/subscribeToAuthCallback"
 import { sendMcpMarketplaceCatalogEvent } from "./mcp/subscribeToMcpMarketplaceCatalog"
 import { sendRelinquishControlEvent } from "./ui/subscribeToRelinquishControl"
+import { handleTaskServiceRequest } from "./task"
+import { BooleanRequest } from "@shared/proto/common"
 
 /*
 https://github.com/microsoft/vscode-webview-ui-toolkit-samples/blob/main/default/weather-webview/src/providers/WeatherViewProvider.ts
@@ -51,10 +54,11 @@ https://github.com/KumarVariable/vscode-extension-sidebar-html/blob/master/src/c
 */
 
 export class Controller {
-	readonly id: string = uuidv4()
+	readonly id: string
 	private postMessage: (message: ExtensionMessage) => Thenable<boolean> | undefined
 
 	private disposables: vscode.Disposable[] = []
+	private mode: "plan" | "act" = "plan" // In-memory plan/act mode state
 	task?: Task
 	workspaceTracker: WorkspaceTracker
 	mcpHub: McpHub
@@ -65,7 +69,9 @@ export class Controller {
 		readonly context: vscode.ExtensionContext,
 		private readonly outputChannel: vscode.OutputChannel,
 		postMessage: (message: ExtensionMessage) => Thenable<boolean> | undefined,
+		id: string,
 	) {
+		this.id = id
 		this.outputChannel.appendLine("ClineProvider instantiated")
 		this.postMessage = postMessage
 
@@ -122,7 +128,7 @@ export class Controller {
 		}
 	}
 
-	async setUserInfo(info?: { displayName: string | null; email: string | null; photoURL: string | null }) {
+	async setUserInfo(info?: UserInfo) {
 		await updateGlobalState(this.context, "userInfo", info)
 	}
 
@@ -132,7 +138,7 @@ export class Controller {
 			apiConfiguration,
 			autoApprovalSettings,
 			browserSettings,
-			chatSettings,
+			chatSettings: storedChatSettings,
 			shellIntegrationTimeout,
 			terminalReuseEnabled,
 			terminalOutputLineLimit,
@@ -142,6 +148,12 @@ export class Controller {
 			taskHistory,
 		} = await getAllExtensionState(this.context)
 
+		// Reconstruct ChatSettings with in-memory mode and stored preferences
+		const chatSettings: ChatSettings = {
+			...storedChatSettings, // Spread stored preferences (preferredLanguage, openAIReasoningEffort)
+			mode: this.mode, // Use in-memory mode (override any stored mode)
+		}
+
 		const NEW_USER_TASK_COUNT_THRESHOLD = 10
 
 		// Check if the user has completed enough tasks to no longer be considered a "new user"
@@ -202,48 +214,10 @@ export class Controller {
 	 */
 	async handleWebviewMessage(message: WebviewMessage) {
 		switch (message.type) {
-			case "authStateChanged":
-				await this.setUserInfo(message.user || undefined)
-				await this.postStateToWebview()
-				break
-
-			case "fetchUserCreditsData": {
-				await this.fetchUserCreditsData()
-				break
-			}
 			case "fetchMcpMarketplace": {
 				await this.fetchMcpMarketplace(message.bool)
 				break
 			}
-
-			// telemetry
-			case "telemetrySetting": {
-				if (message.telemetrySetting) {
-					await this.updateTelemetrySetting(message.telemetrySetting)
-				}
-				await this.postStateToWebview()
-				break
-			}
-
-			case "clearAllTaskHistory": {
-				const answer = await vscode.window.showWarningMessage(
-					"What would you like to delete?",
-					{ modal: true },
-					"Delete All Except Favorites",
-					"Delete Everything",
-					"Cancel",
-				)
-
-				if (answer === "Delete All Except Favorites") {
-					await this.deleteNonFavoriteTaskHistory()
-					await this.postStateToWebview()
-				} else if (answer === "Delete Everything") {
-					await this.deleteAllTaskHistory()
-					await this.postStateToWebview()
-				}
-				sendRelinquishControlEvent()
-				break
-			}
 			case "grpc_request": {
 				if (message.grpc_request) {
 					await handleGrpcRequest(this, message.grpc_request)
@@ -271,6 +245,9 @@ export class Controller {
 	async togglePlanActModeWithChatSettings(chatSettings: ChatSettings, chatContent?: ChatContent): Promise<boolean> {
 		const didSwitchToActMode = chatSettings.mode === "act"
 
+		// Store mode in-memory only
+		this.mode = chatSettings.mode
+
 		// Capture mode switch telemetry | Capture regardless of if we know the taskId
 		telemetryService.captureModeSwitch(this.task?.taskId ?? "0", chatSettings.mode)
 
@@ -440,7 +417,9 @@ export class Controller {
 			}
 		}
 
-		await updateWorkspaceState(this.context, "chatSettings", chatSettings)
+		// Save only non-mode properties to workspace storage
+		const { mode, ...persistentChatSettings }: { mode: string } & StoredChatSettings = chatSettings
+		await updateWorkspaceState(this.context, "chatSettings", persistentChatSettings)
 		await this.postStateToWebview()
 
 		if (this.task) {
@@ -494,20 +473,6 @@ export class Controller {
 		}
 	}
 
-	// Account
-
-	async fetchUserCreditsData() {
-		try {
-			await Promise.all([
-				this.accountService?.fetchBalance(),
-				this.accountService?.fetchUsageTransactions(),
-				this.accountService?.fetchPaymentTransactions(),
-			])
-		} catch (error) {
-			console.error("Failed to fetch user credits data:", error)
-		}
-	}
-
 	// Auth
 
 	public async validateAuthState(state: string | null): Promise<boolean> {
@@ -850,110 +815,6 @@ export class Controller {
 		await downloadTask(historyItem.ts, apiConversationHistory)
 	}
 
-	async deleteAllTaskHistory() {
-		await this.clearTask()
-		await updateGlobalState(this.context, "taskHistory", undefined)
-		try {
-			// Remove all contents of tasks directory
-			const taskDirPath = path.join(this.context.globalStorageUri.fsPath, "tasks")
-			if (await fileExistsAtPath(taskDirPath)) {
-				await fs.rm(taskDirPath, { recursive: true, force: true })
-			}
-			// Remove checkpoints directory contents
-			const checkpointsDirPath = path.join(this.context.globalStorageUri.fsPath, "checkpoints")
-			if (await fileExistsAtPath(checkpointsDirPath)) {
-				await fs.rm(checkpointsDirPath, { recursive: true, force: true })
-			}
-		} catch (error) {
-			vscode.window.showErrorMessage(
-				`Encountered error while deleting task history, there may be some files left behind. Error: ${error instanceof Error ? error.message : String(error)}`,
-			)
-		}
-		// await this.postStateToWebview()
-	}
-
-	async deleteNonFavoriteTaskHistory() {
-		await this.clearTask()
-
-		const taskHistory = ((await getGlobalState(this.context, "taskHistory")) as HistoryItem[]) || []
-		const favoritedTasks = taskHistory.filter((task) => task.isFavorited === true)
-
-		// If user has no favorited tasks, show a warning message
-		if (favoritedTasks.length === 0) {
-			vscode.window.showWarningMessage("No favorited tasks found. Please favorite tasks before using this option.")
-			await this.postStateToWebview()
-			return
-		}
-
-		await updateGlobalState(this.context, "taskHistory", favoritedTasks)
-
-		// Delete non-favorited task directories
-		try {
-			const preserveTaskIds = favoritedTasks.map((task) => task.id)
-			const taskDirPath = path.join(this.context.globalStorageUri.fsPath, "tasks")
-
-			if (await fileExistsAtPath(taskDirPath)) {
-				const taskDirs = await fs.readdir(taskDirPath)
-				for (const taskDir of taskDirs) {
-					if (!preserveTaskIds.includes(taskDir)) {
-						await fs.rm(path.join(taskDirPath, taskDir), { recursive: true, force: true })
-					}
-				}
-			}
-		} catch (error) {
-			vscode.window.showErrorMessage(
-				`Error deleting task history: ${error instanceof Error ? error.message : String(error)}`,
-			)
-		}
-
-		await this.postStateToWebview()
-	}
-
-	async deleteTaskWithId(id: string) {
-		console.info("deleteTaskWithId: ", id)
-
-		try {
-			if (id === this.task?.taskId) {
-				await this.clearTask()
-				console.debug("cleared task")
-			}
-
-			const {
-				taskDirPath,
-				apiConversationHistoryFilePath,
-				uiMessagesFilePath,
-				contextHistoryFilePath,
-				taskMetadataFilePath,
-			} = await this.getTaskWithId(id)
-			const legacyMessagesFilePath = path.join(taskDirPath, "claude_messages.json")
-			const updatedTaskHistory = await this.deleteTaskFromState(id)
-
-			// Delete the task files
-			for (const filePath of [
-				apiConversationHistoryFilePath,
-				uiMessagesFilePath,
-				contextHistoryFilePath,
-				taskMetadataFilePath,
-				legacyMessagesFilePath,
-			]) {
-				const fileExists = await fileExistsAtPath(filePath)
-				if (fileExists) {
-					await fs.unlink(filePath)
-				}
-			}
-
-			await fs.rmdir(taskDirPath) // succeeds if the dir is empty
-
-			if (updatedTaskHistory.length === 0) {
-				await this.deleteAllTaskHistory()
-			}
-		} catch (error) {
-			console.debug(`Error deleting task:`, error)
-		}
-
-		await this.postStateToWebview()
-	}
-
 	async deleteTaskFromState(id: string) {
 		// Remove the task from history
 		const taskHistory = ((await getGlobalState(this.context, "taskHistory")) as HistoryItem[] | undefined) || []
@@ -968,7 +829,7 @@ export class Controller {
 
 	async postStateToWebview() {
 		const state = await this.getStateToPostToWebview()
-		await sendStateUpdate(state)
+		await sendStateUpdate(this.id, state)
 	}
 
 	async getStateToPostToWebview(): Promise<ExtensionState> {
@@ -978,7 +839,7 @@ export class Controller {
 			taskHistory,
 			autoApprovalSettings,
 			browserSettings,
-			chatSettings,
+			chatSettings: storedChatSettings,
 			userInfo,
 			mcpMarketplaceEnabled,
 			mcpRichDisplayEnabled,
@@ -995,6 +856,12 @@ export class Controller {
 			terminalOutputLineLimit,
 		} = await getAllExtensionState(this.context)
 
+		// Reconstruct ChatSettings with in-memory mode and stored preferences
+		const chatSettings: ChatSettings = {
+			...storedChatSettings, // Spread stored preferences (preferredLanguage, openAIReasoningEffort)
+			mode: this.mode, // Use in-memory mode (override any stored mode)
+		}
+
 		const localClineRulesToggles =
 			((await getWorkspaceState(this.context, "localClineRulesToggles")) as ClineRulesToggles) || {}
 

src/core/controller/models/updateApiConfigurationProto.ts

@@ -21,9 +21,15 @@ export async function updateApiConfigurationProto(
 			throw new Error("API configuration is required")
 		}
 
+		console.log("[updateApiConfigurationProto] Received request with provider:", request.apiConfiguration.apiProvider)
+		console.log("[updateApiConfigurationProto] Full proto config:", request.apiConfiguration)
+
 		// Convert proto ApiConfiguration to application ApiConfiguration
 		const appApiConfiguration = convertProtoToApiConfiguration(request.apiConfiguration)
 
+		console.log("[updateApiConfigurationProto] Converted to app config with provider:", appApiConfiguration.apiProvider)
+		console.log("[updateApiConfigurationProto] Full app config:", appApiConfiguration)
+
 		// Update the API configuration in storage
 		await updateApiConfiguration(controller.context, appApiConfiguration)
 

src/core/controller/state/subscribeToState.ts

@@ -3,8 +3,8 @@ import { Controller } from "../index"
 import { EmptyRequest } from "../../../shared/proto/common"
 import { StreamingResponseHandler, getRequestRegistry } from "../grpc-handler"
 
-// Keep track of active state subscriptions
-const activeStateSubscriptions = new Set<StreamingResponseHandler>()
+// Keep track of active state subscriptions by controller ID
+const activeStateSubscriptions = new Map<string, StreamingResponseHandler>()
 
 /**
  * Subscribe to state updates
@@ -19,23 +19,25 @@ export async function subscribeToState(
 	responseStream: StreamingResponseHandler,
 	requestId?: string,
 ): Promise<void> {
+	const controllerId = controller.id
+
 	// Send the initial state
 	const initialState = await controller.getStateToPostToWebview()
 	const initialStateJson = JSON.stringify(initialState)
 
-	console.log("[DEBUG] set up state subscription")
+	console.log(`[DEBUG] set up state subscription for controller ${controllerId}`)
 
 	await responseStream({
 		stateJson: initialStateJson,
 	})
 
-	// Add this subscription to the active subscriptions
-	activeStateSubscriptions.add(responseStream)
+	// Add this subscription to the active subscriptions with the controller ID
+	activeStateSubscriptions.set(controllerId, responseStream)
 
 	// Register cleanup when the connection is closed
 	const cleanup = () => {
-		activeStateSubscriptions.delete(responseStream)
-		console.log("[DEBUG] Cleaned up state subscription")
+		activeStateSubscriptions.delete(controllerId)
+		console.log(`[DEBUG] Cleaned up state subscription for controller ${controllerId}`)
 	}
 
 	// Register the cleanup function with the request registry if we have a requestId
@@ -45,30 +47,31 @@ export async function subscribeToState(
 }
 
 /**
- * Send a state update to all active subscribers
+ * Send a state update to a specific controller's subscription
+ * @param controllerId The ID of the controller to send the state to
  * @param state The state to send
  */
-export async function sendStateUpdate(state: any): Promise<void> {
-	const stateJson = JSON.stringify(state)
+export async function sendStateUpdate(controllerId: string, state: any): Promise<void> {
+	// Get the subscription for this specific controller
+	const responseStream = activeStateSubscriptions.get(controllerId)
 
-	// Send the update to all active subscribers
-	const promises = Array.from(activeStateSubscriptions).map(async (responseStream) => {
-		try {
-			// The issue might be that we're not properly formatting the response
-			// Let's ensure we're sending a properly formatted State message
-			await responseStream(
-				{
-					stateJson,
-				},
-				false, // Not the last message
-			)
-			console.log("[DEBUG] sending followup state", stateJson.length, "chars")
-		} catch (error) {
-			console.error("Error sending state update:", error)
-			// Remove the subscription if there was an error
-			activeStateSubscriptions.delete(responseStream)
-		}
-	})
+	if (!responseStream) {
+		console.log(`[DEBUG] No active state subscription for controller ${controllerId}`)
+		return
+	}
 
-	await Promise.all(promises)
+	try {
+		const stateJson = JSON.stringify(state)
+		await responseStream(
+			{
+				stateJson,
+			},
+			false, // Not the last message
+		)
+		console.log(`[DEBUG] sending followup state to controller ${controllerId}`, stateJson.length, "chars")
+	} catch (error) {
+		console.error(`Error sending state update to controller ${controllerId}:`, error)
+		// Remove the subscription if there was an error
+		activeStateSubscriptions.delete(controllerId)
+	}
 }

src/core/controller/state/updateSettings.ts

@@ -58,7 +58,7 @@ export async function updateSettings(controller: Controller, request: UpdateSett
 		// Update chat settings
 		if (request.chatSettings) {
 			const chatSettings = convertProtoChatSettingsToChatSettings(request.chatSettings)
-			await controller.context.globalState.update("chatSettings", chatSettings)
+			await controller.context.workspaceState.update("chatSettings", chatSettings)
 			if (controller.task) {
 				controller.task.chatSettings = chatSettings
 			}

src/core/controller/state/updateTelemetrySetting.ts

@@ -0,0 +1,16 @@
+import { Controller } from ".."
+import { Empty } from "../../../shared/proto/common"
+import { TelemetrySettingRequest } from "../../../shared/proto/state"
+import { convertProtoTelemetrySettingToDomain } from "../../../shared/proto-conversions/state/telemetry-setting-conversion"
+
+/**
+ * Updates the telemetry setting
+ * @param controller The controller instance
+ * @param request The telemetry setting request
+ * @returns Empty response
+ */
+export async function updateTelemetrySetting(controller: Controller, request: TelemetrySettingRequest): Promise<Empty> {
+	const telemetrySetting = convertProtoTelemetrySettingToDomain(request.setting)
+	await controller.updateTelemetrySetting(telemetrySetting)
+	return Empty.create()
+}

src/core/controller/task/deleteAllTaskHistory.ts

@@ -0,0 +1,114 @@
+import path from "path"
+import fs from "fs/promises"
+import { Controller } from ".."
+import { DeleteAllTaskHistoryCount } from "../../../shared/proto/task"
+import { BooleanRequest } from "../../../shared/proto/common"
+import { getGlobalState, updateGlobalState } from "../../storage/state"
+import { fileExistsAtPath } from "../../../utils/fs"
+import vscode from "vscode"
+
+/**
+ * Deletes all task history, with an option to preserve favorites
+ * @param controller The controller instance
+ * @param request Request with option to preserve favorites
+ * @returns Results with count of deleted tasks
+ */
+export async function deleteAllTaskHistory(controller: Controller, request: BooleanRequest): Promise<DeleteAllTaskHistoryCount> {
+	try {
+		// Clear current task first
+		await controller.clearTask()
+
+		// Get existing task history
+		const taskHistory = ((await getGlobalState(controller.context, "taskHistory")) as any[]) || []
+		const totalTasks = taskHistory.length
+
+		// If preserving favorites, filter out non-favorites
+		if (request.value) {
+			const favoritedTasks = taskHistory.filter((task) => task.isFavorited === true)
+
+			// If there are favorited tasks, update state
+			if (favoritedTasks.length > 0) {
+				await updateGlobalState(controller.context, "taskHistory", favoritedTasks)
+
+				// Delete non-favorited task directories
+				const preserveTaskIds = favoritedTasks.map((task) => task.id)
+				await cleanupTaskFiles(controller, preserveTaskIds)
+
+				// Update webview
+				try {
+					await controller.postStateToWebview()
+				} catch (webviewErr) {
+					console.error("Error posting to webview:", webviewErr)
+				}
+
+				return DeleteAllTaskHistoryCount.create({
+					tasksDeleted: totalTasks - favoritedTasks.length,
+				})
+			} else {
+				return DeleteAllTaskHistoryCount.create({
+					tasksDeleted: 0,
+				})
+			}
+		}
+
+		// Delete everything (not preserving favorites)
+		await updateGlobalState(controller.context, "taskHistory", undefined)
+
+		try {
+			// Remove all contents of tasks directory
+			const taskDirPath = path.join(controller.context.globalStorageUri.fsPath, "tasks")
+			if (await fileExistsAtPath(taskDirPath)) {
+				await fs.rm(taskDirPath, { recursive: true, force: true })
+			}
+
+			// Remove checkpoints directory contents
+			const checkpointsDirPath = path.join(controller.context.globalStorageUri.fsPath, "checkpoints")
+			if (await fileExistsAtPath(checkpointsDirPath)) {
+				await fs.rm(checkpointsDirPath, { recursive: true, force: true })
+			}
+		} catch (error) {
+			vscode.window.showErrorMessage(
+				`Encountered error while deleting task history, there may be some files left behind. Error: ${error instanceof Error ? error.message : String(error)}`,
+			)
+		}
+
+		// Update webview
+		try {
+			await controller.postStateToWebview()
+		} catch (webviewErr) {
+			console.error("Error posting to webview:", webviewErr)
+		}
+
+		return DeleteAllTaskHistoryCount.create({
+			tasksDeleted: totalTasks,
+		})
+	} catch (error) {
+		console.error("Error in deleteAllTaskHistory:", error)
+		throw error
+	}
+}
+
+/**
+ * Helper function to cleanup task files while preserving specified tasks
+ */
+async function cleanupTaskFiles(controller: Controller, preserveTaskIds: string[]) {
+	const taskDirPath = path.join(controller.context.globalStorageUri.fsPath, "tasks")
+
+	try {
+		if (await fileExistsAtPath(taskDirPath)) {
+			const taskDirs = await fs.readdir(taskDirPath)
+			console.debug(`[cleanupTaskFiles] Found ${taskDirs.length} task directories`)
+
+			// Delete only non-preserved task directories
+			for (const dir of taskDirs) {
+				if (!preserveTaskIds.includes(dir)) {
+					await fs.rm(path.join(taskDirPath, dir), { recursive: true, force: true })
+				}
+			}
+		}
+	} catch (error) {
+		console.error("Error cleaning up task files:", error)
+	}
+
+	return true
+}

src/core/controller/task/deleteTasksWithIds.ts

@@ -1,6 +1,9 @@
+import path from "path"
+import fs from "fs/promises"
 import { Controller } from ".."
-import { Empty, StringArrayRequest } from "../../../shared/proto/common"
+import { Empty, StringArrayRequest, BooleanRequest } from "../../../shared/proto/common"
 import { TaskMethodHandler } from "./index"
+import { fileExistsAtPath } from "../../../utils/fs"
 
 /**
  * Deletes tasks with the specified IDs
@@ -17,7 +20,72 @@ export const deleteTasksWithIds: TaskMethodHandler = async (
 		throw new Error("Missing task IDs")
 	}
 
-	await Promise.all(request.value.map((value) => controller.deleteTaskWithId(value)))
+	for (const id of request.value) {
+		await deleteTaskWithId(controller, id)
+	}
 
 	return Empty.create()
 }
+
+/**
+ * Deletes a single task with the specified ID
+ * @param controller The controller instance
+ * @param id The task ID to delete
+ */
+async function deleteTaskWithId(controller: Controller, id: string): Promise<void> {
+	console.info("deleteTaskWithId: ", id)
+
+	try {
+		// Clear current task if it matches the ID being deleted
+		if (id === controller.task?.taskId) {
+			await controller.clearTask()
+			console.debug("cleared task")
+		}
+
+		// Get task file paths
+		const { taskDirPath, apiConversationHistoryFilePath, uiMessagesFilePath, contextHistoryFilePath, taskMetadataFilePath } =
+			await controller.getTaskWithId(id)
+
+		// Remove task from state
+		const updatedTaskHistory = await controller.deleteTaskFromState(id)
+
+		// Delete the task files
+		for (const filePath of [
+			apiConversationHistoryFilePath,
+			uiMessagesFilePath,
+			contextHistoryFilePath,
+			taskMetadataFilePath,
+		]) {
+			const fileExists = await fileExistsAtPath(filePath)
+			if (fileExists) {
+				await fs.unlink(filePath)
+			}
+		}
+
+		// Remove empty task directory
+		try {
+			await fs.rmdir(taskDirPath) // succeeds if the dir is empty
+		} catch (error) {
+			console.debug("Could not remove task directory (may not be empty):", error)
+		}
+
+		// If no tasks remain, clean up everything
+		if (updatedTaskHistory.length === 0) {
+			const taskDirPath = path.join(controller.context.globalStorageUri.fsPath, "tasks")
+			const checkpointsDirPath = path.join(controller.context.globalStorageUri.fsPath, "checkpoints")
+
+			if (await fileExistsAtPath(taskDirPath)) {
+				await fs.rm(taskDirPath, { recursive: true, force: true })
+			}
+			if (await fileExistsAtPath(checkpointsDirPath)) {
+				await fs.rm(checkpointsDirPath, { recursive: true, force: true })
+			}
+		}
+	} catch (error) {
+		console.debug(`Error deleting task ${id}:`, error)
+		throw error // Re-throw to let caller handle the error
+	}
+
+	// Update webview state
+	await controller.postStateToWebview()
+}

src/core/controller/ui/getWebviewHtml.ts

@@ -0,0 +1,16 @@
+import type { Controller } from "../index"
+import { EmptyRequest, Empty, String } from "@shared/proto/common"
+import * as hostProviders from "@hosts/host-providers"
+import { WebviewProviderType } from "@/shared/webview/types"
+
+/**
+ * Initialize webview when it launches
+ * @param controller The controller instance
+ * @param request The empty request
+ * @returns Empty response
+ */
+export async function getWebviewHtml(_controller: Controller, _: EmptyRequest): Promise<String> {
+	const webviewProvider = hostProviders.createWebviewProvider(WebviewProviderType.SIDEBAR)
+
+	return Promise.resolve(String.create({ value: webviewProvider.getHtmlContent() }))
+}

src/core/controller/ui/openUrl.ts

@@ -0,0 +1,20 @@
+import type { Controller } from "../index"
+import type { StringRequest } from "../../../shared/proto/common"
+import { Empty } from "../../../shared/proto/common"
+import { openUrlInBrowser } from "../../../utils/github-url-utils"
+
+/**
+ * Opens a URL in the default browser
+ * @param controller The controller instance
+ * @param request The URL to open
+ * @returns Empty response
+ */
+export async function openUrl(controller: Controller, request: StringRequest): Promise<Empty> {
+	try {
+		await openUrlInBrowser(request.value)
+		return Empty.create({})
+	} catch (error) {
+		console.error(`Failed to open URL: ${error}`)
+		throw error
+	}
+}

src/core/controller/ui/subscribeToMcpButtonClicked.ts

@@ -1,33 +1,33 @@
 import { Controller } from "../index"
 import { Empty } from "@shared/proto/common"
-import { WebviewProviderType, WebviewProviderTypeRequest } from "@shared/proto/ui"
+import { EmptyRequest } from "@shared/proto/common"
 import { StreamingResponseHandler, getRequestRegistry } from "../grpc-handler"
 
-// Track subscriptions with their provider type
-const mcpButtonClickedSubscriptions = new Map<StreamingResponseHandler, WebviewProviderType>()
+// Keep track of active mcpButtonClicked subscriptions by controller ID
+const activeMcpButtonClickedSubscriptions = new Map<string, StreamingResponseHandler>()
 
 /**
  * Subscribe to mcpButtonClicked events
  * @param controller The controller instance
- * @param request The webview provider type request
+ * @param request The empty request
  * @param responseStream The streaming response handler
  * @param requestId The ID of the request (passed by the gRPC handler)
  */
 export async function subscribeToMcpButtonClicked(
-	_controller: Controller,
-	request: WebviewProviderTypeRequest,
+	controller: Controller,
+	_request: EmptyRequest,
 	responseStream: StreamingResponseHandler,
 	requestId?: string,
 ): Promise<void> {
-	const providerType = request.providerType
-	console.log(`[DEBUG] set up mcpButtonClicked subscription for ${WebviewProviderType[providerType]} webview`)
+	const controllerId = controller.id
+	console.log(`[DEBUG] set up mcpButtonClicked subscription for controller ${controllerId}`)
 
-	// Store the subscription with its provider type
-	mcpButtonClickedSubscriptions.set(responseStream, providerType)
+	// Add this subscription to the active subscriptions with the controller ID
+	activeMcpButtonClickedSubscriptions.set(controllerId, responseStream)
 
 	// Register cleanup when the connection is closed
 	const cleanup = () => {
-		mcpButtonClickedSubscriptions.delete(responseStream)
+		activeMcpButtonClickedSubscriptions.delete(controllerId)
 	}
 
 	// Register the cleanup function with the request registry if we have a requestId
@@ -37,26 +37,27 @@ export async function subscribeToMcpButtonClicked(
 }
 
 /**
- * Send a mcpButtonClicked event to active subscribers based on webview type
- * @param webviewType The type of webview that triggered the event (SIDEBAR or TAB)
+ * Send a mcpButtonClicked event to a specific controller's subscription
+ * @param controllerId The ID of the controller to send the event to
  */
-export async function sendMcpButtonClickedEvent(webviewType?: WebviewProviderType): Promise<void> {
-	const event = Empty.create({})
+export async function sendMcpButtonClickedEvent(controllerId: string): Promise<void> {
+	// Get the subscription for this specific controller
+	const responseStream = activeMcpButtonClickedSubscriptions.get(controllerId)
 
-	// Process all subscriptions, filtering based on the source
-	const promises = Array.from(mcpButtonClickedSubscriptions.entries()).map(async ([responseStream, providerType]) => {
-		// Only send to subscribers of the same type as the event source
-		if (webviewType !== providerType) {
-			return // Skip subscribers of different types
-		}
+	if (!responseStream) {
+		console.error(`[DEBUG] No active subscription for controller ${controllerId}`)
+		return
+	}
 
-		try {
-			await responseStream(event, false)
-		} catch (error) {
-			console.error(`Error sending mcpButtonClicked event to ${WebviewProviderType[providerType]}:`, error)
-			mcpButtonClickedSubscriptions.delete(responseStream)
-		}
-	})
-
-	await Promise.all(promises)
+	try {
+		const event = Empty.create({})
+		await responseStream(
+			event,
+			false, // Not the last message
+		)
+	} catch (error) {
+		console.error(`Error sending mcpButtonClicked event to controller ${controllerId}:`, error)
+		// Remove the subscription if there was an error
+		activeMcpButtonClickedSubscriptions.delete(controllerId)
+	}
 }

src/core/prompts/model_prompts/claude4-experimental.ts

@@ -128,7 +128,10 @@ ${
 					const config = JSON.parse(server.config)
 
 					return (
-						`## ${server.name} (\`${config.command}${config.args && Array.isArray(config.args) ? ` ${config.args.join(" ")}` : ""}\`)` +
+						`## ${server.name}` +
+						(config.command
+							? ` (\`${config.command}${config.args && Array.isArray(config.args) ? ` ${config.args.join(" ")}` : ""}\`)`
+							: "") +
 						(tools ? `\n\n### Available Tools\n${tools}` : "") +
 						(templates ? `\n\n### Resource Templates\n${templates}` : "") +
 						(resources ? `\n\n### Direct Resources\n${resources}` : "")

src/core/prompts/model_prompts/claude4.ts

@@ -470,7 +470,10 @@ ${
                     const config = JSON.parse(server.config)
 
                     return (
-                        `## ${server.name} (\`${config.command}${config.args && Array.isArray(config.args) ? ` ${config.args.join(" ")}` : ""}\`)` +
+						`## ${server.name}` +
+						(config.command
+							? ` (\`${config.command}${config.args && Array.isArray(config.args) ? ` ${config.args.join(" ")}` : ""}\`)`
+							: "") +
                         (tools ? `\n\n### Available Tools\n${tools}` : "") +
                         (templates ? `\n\n### Resource Templates\n${templates}` : "") +
                         (resources ? `\n\n### Direct Resources\n${resources}` : "")

src/core/prompts/system.ts

@@ -466,7 +466,10 @@ ${
 					const config = JSON.parse(server.config)
 
 					return (
-						`## ${server.name} (\`${config.command}${config.args && Array.isArray(config.args) ? ` ${config.args.join(" ")}` : ""}\`)` +
+						`## ${server.name}` +
+						(config.command
+							? ` (\`${config.command}${config.args && Array.isArray(config.args) ? ` ${config.args.join(" ")}` : ""}\`)`
+							: "") +
 						(tools ? `\n\n### Available Tools\n${tools}` : "") +
 						(templates ? `\n\n### Resource Templates\n${templates}` : "") +
 						(resources ? `\n\n### Direct Resources\n${resources}` : "")

src/core/storage/state.ts

@@ -7,7 +7,7 @@ import { ApiConfiguration, ApiProvider, BedrockModelId, ModelInfo } from "@share
 import { HistoryItem } from "@shared/HistoryItem"
 import { AutoApprovalSettings } from "@shared/AutoApprovalSettings"
 import { BrowserSettings } from "@shared/BrowserSettings"
-import { ChatSettings } from "@shared/ChatSettings"
+import { ChatSettings, StoredChatSettings } from "@shared/ChatSettings"
 import { TelemetrySetting } from "@shared/TelemetrySetting"
 import { UserInfo } from "@shared/UserInfo"
 import { ClineRulesToggles } from "@shared/cline-rules"
@@ -173,6 +173,28 @@ export async function migrateCustomInstructionsToGlobalRules(context: vscode.Ext
 	}
 }
 
+export async function cleanupModeFromWorkspaceStorage(context: vscode.ExtensionContext) {
+	try {
+		// Get current chatSettings from workspace storage
+		const chatSettings = (await getWorkspaceState(context, "chatSettings")) as any
+
+		if (chatSettings && typeof chatSettings === "object" && "mode" in chatSettings) {
+			console.log("Cleaning up mode from workspace storage...")
+
+			// Remove mode property from chatSettings
+			const { mode, ...cleanedChatSettings } = chatSettings
+
+			// Save cleaned chatSettings back to workspace storage
+			await updateWorkspaceState(context, "chatSettings", cleanedChatSettings)
+
+			console.log("Successfully removed mode from workspace storage chatSettings")
+		}
+	} catch (error) {
+		console.error("Failed to cleanup mode from workspace storage:", error)
+		// Continue execution - migration failure shouldn't break extension startup
+	}
+}
+
 export async function getAllExtensionState(context: vscode.ExtensionContext) {
 	const [
 		isNewUser,
@@ -360,7 +382,7 @@ export async function getAllExtensionState(context: vscode.ExtensionContext) {
 		previousModeSapAiCoreResourceGroup,
 		previousModeSapAiCoreModelId,
 	] = await Promise.all([
-		getWorkspaceState(context, "chatSettings") as Promise<ChatSettings | undefined>,
+		getWorkspaceState(context, "chatSettings") as Promise<StoredChatSettings | undefined>,
 		getWorkspaceState(context, "apiProvider") as Promise<ApiProvider | undefined>,
 		getWorkspaceState(context, "apiModelId") as Promise<string | undefined>,
 		getWorkspaceState(context, "thinkingBudgetTokens") as Promise<number | undefined>,

src/core/task/TaskState.ts

@@ -34,6 +34,9 @@ export class TaskState {
 	isAwaitingPlanResponse = false
 	didRespondToPlanAskBySwitchingMode = false
 
+	// Context and history
+	conversationHistoryDeletedRange?: [number, number]
+
 	// Tool execution flags
 	didRejectTool = false
 	didAlreadyUseTool = false

src/core/task/message-state.ts

@@ -11,13 +11,14 @@ import * as path from "path"
 import CheckpointTracker from "@integrations/checkpoints/CheckpointTracker"
 import { HistoryItem } from "@/shared/HistoryItem"
 import Anthropic from "@anthropic-ai/sdk"
+import { TaskState } from "./TaskState"
 
 interface MessageStateHandlerParams {
 	context: vscode.ExtensionContext
 	taskId: string
-	conversationHistoryDeletedRange?: [number, number]
 	taskIsFavorited?: boolean
 	updateTaskHistory: (historyItem: HistoryItem) => Promise<HistoryItem[]>
+	taskState: TaskState
 }
 
 const cwd = vscode.workspace.workspaceFolders?.map((folder) => folder.uri.fsPath).at(0) ?? path.join(os.homedir(), "Desktop") // may or may not exist but fs checking existence would immediately ask for permission which would be bad UX, need to come up with a better solution
@@ -25,17 +26,17 @@ const cwd = vscode.workspace.workspaceFolders?.map((folder) => folder.uri.fsPath
 export class MessageStateHandler {
 	private apiConversationHistory: Anthropic.MessageParam[] = []
 	private clineMessages: ClineMessage[] = []
-	private conversationHistoryDeletedRange: [number, number] | undefined
 	private taskIsFavorited: boolean
 	private checkpointTracker: CheckpointTracker | undefined
 	private updateTaskHistory: (historyItem: HistoryItem) => Promise<HistoryItem[]>
 	private context: vscode.ExtensionContext
 	private taskId: string
+	private taskState: TaskState
 
 	constructor(params: MessageStateHandlerParams) {
 		this.context = params.context
 		this.taskId = params.taskId
-		this.conversationHistoryDeletedRange = params.conversationHistoryDeletedRange
+		this.taskState = params.taskState
 		this.taskIsFavorited = params.taskIsFavorited ?? false
 		this.updateTaskHistory = params.updateTaskHistory
 	}
@@ -95,7 +96,7 @@ export class MessageStateHandler {
 				size: taskDirSize,
 				shadowGitConfigWorkTree: await this.checkpointTracker?.getShadowGitConfigWorkTree(),
 				cwdOnTaskInitialization: cwd,
-				conversationHistoryDeletedRange: this.conversationHistoryDeletedRange,
+				conversationHistoryDeletedRange: this.taskState.conversationHistoryDeletedRange,
 				isFavorited: this.taskIsFavorited,
 			})
 		} catch (error) {
@@ -117,7 +118,7 @@ export class MessageStateHandler {
 		// these values allow us to reconstruct the conversation history at the time this cline message was created
 		// it's important that apiConversationHistory is initialized before we add cline messages
 		message.conversationHistoryIndex = this.apiConversationHistory.length - 1 // NOTE: this is the index of the last added message which is the user message, and once the clinemessages have been presented we update the apiconversationhistory with the completed assistant message. This means when resetting to a message, we need to +1 this index to get the correct assistant message that this tool use corresponds to
-		message.conversationHistoryDeletedRange = this.conversationHistoryDeletedRange
+		message.conversationHistoryDeletedRange = this.taskState.conversationHistoryDeletedRange
 		this.clineMessages.push(message)
 		await this.saveClineMessagesAndUpdateHistory()
 	}

src/core/webview/VscodeWebviewProvider.ts

@@ -15,14 +15,6 @@ https://github.com/KumarVariable/vscode-extension-sidebar-html/blob/master/src/c
 export class VscodeWebviewProvider extends WebviewProvider implements vscode.WebviewViewProvider {
 	public webview?: vscode.WebviewView | vscode.WebviewPanel
 
-	public static create(
-		context: vscode.ExtensionContext,
-		outputChannel: vscode.OutputChannel,
-		providerType: WebviewProviderType,
-	) {
-		return new VscodeWebviewProvider(context, outputChannel, providerType)
-	}
-
 	constructor(context: vscode.ExtensionContext, outputChannel: vscode.OutputChannel, providerType: WebviewProviderType) {
 		super(context, outputChannel, providerType)
 	}

src/core/webview/index.ts

@@ -28,7 +28,7 @@ export abstract class WebviewProvider {
 		WebviewProvider.activeInstances.add(this)
 		this.clientId = uuidv4()
 		WebviewProvider.clientIdMap.set(this, this.clientId)
-		this.controller = new Controller(context, outputChannel, (message) => this.postMessageToWebview(message))
+		this.controller = new Controller(context, outputChannel, (message) => this.postMessageToWebview(message), this.clientId)
 	}
 
 	// Add a method to get the client ID
@@ -58,6 +58,19 @@ export abstract class WebviewProvider {
 		return findLast(Array.from(this.activeInstances), (instance) => instance.isVisible() === true)
 	}
 
+	public static getActiveInstance(): WebviewProvider | undefined {
+		return Array.from(this.activeInstances).find((instance) => {
+			if (
+				instance.getWebview() &&
+				instance.getWebview().viewType === "claude-dev.TabPanelProvider" &&
+				"active" in instance.getWebview()
+			) {
+				return instance.getWebview().active === true
+			}
+			return false
+		})
+	}
+
 	public static getAllInstances(): WebviewProvider[] {
 		return Array.from(this.activeInstances)
 	}
@@ -187,7 +200,7 @@ export abstract class WebviewProvider {
 				<link href="${codiconsUri}" rel="stylesheet" />
 				<link href="${katexCssUri}" rel="stylesheet" />
 				<meta http-equiv="Content-Security-Policy" content="default-src 'none';
-					connect-src https://*.posthog.com https://*.firebaseauth.com https://*.firebaseio.com https://*.googleapis.com https://*.firebase.com; 
+					connect-src https://*.posthog.com https://*.cline.bot https://*.firebaseauth.com https://*.firebaseio.com https://*.googleapis.com https://*.firebase.com; 
 					font-src ${this.getCspSource()} data:; 
 					style-src ${this.getCspSource()} 'unsafe-inline'; 
 					img-src ${this.getCspSource()} https: data:; 

src/extension.ts

@@ -22,13 +22,18 @@ import { WebviewProviderType as WebviewProviderTypeEnum } from "@shared/proto/ui
 import { WebviewProviderType } from "./shared/webview/types"
 import { sendHistoryButtonClickedEvent } from "./core/controller/ui/subscribeToHistoryButtonClicked"
 import { sendAccountButtonClickedEvent } from "./core/controller/ui/subscribeToAccountButtonClicked"
-import { migratePlanActGlobalToWorkspaceStorage, migrateCustomInstructionsToGlobalRules } from "./core/storage/state"
+import {
+	migratePlanActGlobalToWorkspaceStorage,
+	migrateCustomInstructionsToGlobalRules,
+	cleanupModeFromWorkspaceStorage,
+} from "./core/storage/state"
 
 import { sendFocusChatInputEvent } from "./core/controller/ui/subscribeToFocusChatInput"
 import { FileContextTracker } from "./core/context/context-tracking/FileContextTracker"
 import * as hostProviders from "@hosts/host-providers"
 import { vscodeHostBridgeClient } from "@/hosts/vscode/client/host-grpc-client"
 import { VscodeWebviewProvider } from "./core/webview/VscodeWebviewProvider"
+import { ExtensionContext } from "vscode"
 
 /*
 Built using https://github.com/microsoft/vscode-webview-ui-toolkit
@@ -51,7 +56,7 @@ export async function activate(context: vscode.ExtensionContext) {
 	Logger.initialize(outputChannel)
 	Logger.log("Cline extension activated")
 
-	maybeSetupHostProviders()
+	maybeSetupHostProviders(context)
 
 	// Migrate global storage values to workspace storage (one-time cleanup)
 	await migratePlanActGlobalToWorkspaceStorage(context)
@@ -59,13 +64,16 @@ export async function activate(context: vscode.ExtensionContext) {
 	// Migrate custom instructions to global Cline rules (one-time cleanup)
 	await migrateCustomInstructionsToGlobalRules(context)
 
+	// Clean up mode from workspace storage (one-time cleanup)
+	await cleanupModeFromWorkspaceStorage(context)
+
 	// Clean up orphaned file context warnings (startup cleanup)
 	await FileContextTracker.cleanupOrphanedWarnings(context)
 
 	// Version checking for autoupdate notification
 	const currentVersion = context.extension.packageJSON.version
 	const previousVersion = context.globalState.get<string>("clineVersion")
-	const sidebarWebview = hostProviders.createWebviewProvider(context, outputChannel, WebviewProviderType.SIDEBAR)
+	const sidebarWebview = hostProviders.createWebviewProvider(WebviewProviderType.SIDEBAR)
 
 	// Initialize test mode and add disposables to context
 	context.subscriptions.push(...initializeTestMode(context, sidebarWebview))
@@ -141,12 +149,26 @@ export async function activate(context: vscode.ExtensionContext) {
 	context.subscriptions.push(
 		vscode.commands.registerCommand("cline.mcpButtonClicked", (webview: any) => {
 			console.log("[DEBUG] mcpButtonClicked", webview)
-			// Pass the webview type to the event sender
-			const isSidebar = !webview
-			const webviewType = isSidebar ? WebviewProviderTypeEnum.SIDEBAR : WebviewProviderTypeEnum.TAB
 
-			// Will send to appropriate subscribers based on the source webview type
-			sendMcpButtonClickedEvent(webviewType)
+			const activeInstance = WebviewProvider.getActiveInstance()
+			const isSidebar = !webview
+
+			if (isSidebar) {
+				const sidebarInstance = WebviewProvider.getSidebarInstance()
+				const sidebarInstanceId = sidebarInstance?.getClientId()
+				if (sidebarInstanceId) {
+					sendMcpButtonClickedEvent(sidebarInstanceId)
+				} else {
+					console.error("[DEBUG] No sidebar instance found, cannot send MCP button event")
+				}
+			} else {
+				const activeInstanceId = activeInstance?.getClientId()
+				if (activeInstanceId) {
+					sendMcpButtonClickedEvent(activeInstanceId)
+				} else {
+					console.error("[DEBUG] No active instance found, cannot send MCP button event")
+				}
+			}
 		}),
 	)
 
@@ -154,7 +176,7 @@ export async function activate(context: vscode.ExtensionContext) {
 		Logger.log("Opening Cline in new tab")
 		// (this example uses webviewProvider activation event which is necessary to deserialize cached webview, but since we use retainContextWhenHidden, we don't need to use that event)
 		// https://github.com/microsoft/vscode-extension-samples/blob/main/webview-sample/src/extension.ts
-		const tabWebview = hostProviders.createWebviewProvider(context, outputChannel, WebviewProviderType.TAB)
+		const tabWebview = hostProviders.createWebviewProvider(WebviewProviderType.TAB)
 		//const column = vscode.window.activeTextEditor ? vscode.window.activeTextEditor.viewColumn : undefined
 		const lastCol = Math.max(...vscode.window.visibleTextEditors.map((editor) => editor.viewColumn || 0))
 
@@ -628,7 +650,7 @@ export async function activate(context: vscode.ExtensionContext) {
 			} else {
 				// Create a temporary controller just for this operation
 				const outputChannel = vscode.window.createOutputChannel("Cline Commit Generator")
-				const tempController = new Controller(context, outputChannel, () => Promise.resolve(true))
+				const tempController = new Controller(context, outputChannel, () => Promise.resolve(true), uuidv4())
 
 				await tempController.generateGitCommitMessage()
 				outputChannel.dispose()
@@ -639,10 +661,13 @@ export async function activate(context: vscode.ExtensionContext) {
 	return createClineAPI(outputChannel, sidebarWebview.controller)
 }
 
-function maybeSetupHostProviders() {
+function maybeSetupHostProviders(context: ExtensionContext) {
 	if (!hostProviders.isSetup) {
 		console.log("Setting up vscode host providers...")
-		hostProviders.initializeHostProviders(VscodeWebviewProvider.create, vscodeHostBridgeClient)
+		const createWebview = function (type: WebviewProviderType) {
+			return new VscodeWebviewProvider(context, outputChannel, type)
+		}
+		hostProviders.initializeHostProviders(createWebview, vscodeHostBridgeClient)
 	}
 }