+ {markdoc &&
diff --git a/packages/kilo-docs/pages/code-with-ai/agents/auto-model.md b/packages/kilo-docs/pages/code-with-ai/agents/auto-model.md
index ce89c2efb24..94a48c21970 100644
--- a/packages/kilo-docs/pages/code-with-ai/agents/auto-model.md
+++ b/packages/kilo-docs/pages/code-with-ai/agents/auto-model.md
@@ -45,28 +45,28 @@ That's it. No configuration needed.
## Auto Balanced
-`kilo-auto/balanced` follows the same mode-based routing structure as Frontier but uses more cost-effective models — Kimi K2.5 for reasoning-heavy modes and Minimax M2.5 (Free) for implementation modes.
+`kilo-auto/balanced` follows the same mode-based routing structure as Frontier but uses more cost-effective models — Kimi K2.5 for reasoning-heavy modes and Minimax M2.7 for implementation modes.
### Mode-to-Model Mapping
-| Mode | Model Used | Best For |
-| -------------- | ------------------- | ---------------------------- |
-| `architect` | Kimi K2.5 | System design, planning |
-| `orchestrator` | Kimi K2.5 | Multi-step task coordination |
-| `ask` | Kimi K2.5 | Questions, explanations |
-| `plan` | Kimi K2.5 | Planning, reasoning |
-| `general` | Kimi K2.5 | General assistance |
-| `debug` | Kimi K2.5 | Debugging and fixing issues |
-| `code` | Minimax M2.5 (Free) | Writing and editing code |
-| `build` | Minimax M2.5 (Free) | Implementation tasks |
-| `explore` | Minimax M2.5 (Free) | Codebase exploration |
+| Mode | Model Used | Best For |
+| -------------- | ------------ | ---------------------------- |
+| `architect` | Kimi K2.5 | System design, planning |
+| `orchestrator` | Kimi K2.5 | Multi-step task coordination |
+| `ask` | Kimi K2.5 | Questions, explanations |
+| `plan` | Kimi K2.5 | Planning, reasoning |
+| `general` | Kimi K2.5 | General assistance |
+| `debug` | Kimi K2.5 | Debugging and fixing issues |
+| `code` | Minimax M2.7 | Writing and editing code |
+| `build` | Minimax M2.7 | Implementation tasks |
+| `explore` | Minimax M2.7 | Codebase exploration |
**Planning and reasoning tasks** use Kimi K2.5, a strong open-weight reasoning model from Moonshot AI.
-**Implementation tasks** use Minimax M2.5, which provides fast, capable code generation at a fraction of frontier model costs.
+**Implementation tasks** use Minimax M2.7, which provides fast, capable code generation at a fraction of frontier model costs.
{% callout type="info" title="Image support" %}
-Auto Balanced does not support image inputs, since Minimax M2.5 does not have vision capabilities.
+Auto Balanced does not support image inputs, since Minimax M2.7 does not have vision capabilities.
{% /callout %}
## Benefits
diff --git a/packages/kilo-docs/pages/gateway/models-and-providers.md b/packages/kilo-docs/pages/gateway/models-and-providers.md
index aa5c7cf8c01..1a902b2c392 100644
--- a/packages/kilo-docs/pages/gateway/models-and-providers.md
+++ b/packages/kilo-docs/pages/gateway/models-and-providers.md
@@ -82,11 +82,11 @@ Routes to the most capable paid models optimizing for cost, performance, and cap
Follows the same mode-based routing as Frontier but uses more cost-effective models.
-| Mode | Resolved Model |
-| -------------------------------------------------------------- | --------------------------- |
-| `plan`, `general`, `architect`, `orchestrator`, `ask`, `debug` | `moonshotai/kimi-k2.5` |
-| `build`, `explore`, `code` | `minimax/minimax-m2.5:free` |
-| Default (no mode specified) | `minimax/minimax-m2.5:free` |
+| Mode | Resolved Model |
+| -------------------------------------------------------------- | ---------------------- |
+| `plan`, `general`, `architect`, `orchestrator`, `ask`, `debug` | `moonshotai/kimi-k2.5` |
+| `build`, `explore`, `code` | `minimax/minimax-m2.7` |
+| Default (no mode specified) | `minimax/minimax-m2.7` |
### `kilo-auto/free`
diff --git a/packages/kilo-docs/pages/getting-started/byok.md b/packages/kilo-docs/pages/getting-started/byok.md
index 3df9aff13e1..24ed48bf863 100644
--- a/packages/kilo-docs/pages/getting-started/byok.md
+++ b/packages/kilo-docs/pages/getting-started/byok.md
@@ -20,10 +20,11 @@ Kilo Gateway currently supports BYOK keys for these providers:
- Anthropic
- AWS Bedrock
-- OpenAI
- Google AI Studio
+- Inception
- Minimax
- Mistral AI
+- OpenAI
- xAI
- Z.AI
diff --git a/packages/kilo-docs/pages/getting-started/installing.md b/packages/kilo-docs/pages/getting-started/installing.md
index f3693d1b7c6..e63c864adb5 100644
--- a/packages/kilo-docs/pages/getting-started/installing.md
+++ b/packages/kilo-docs/pages/getting-started/installing.md
@@ -14,54 +14,16 @@ Get started with Kilo Code by installing it on your preferred platform. Choose y
## VS Code Extension
-{% partial file="install-vscode.md" /%}
-
-{% /tab %}
-{% tab label="VS Code (Preview)" %}
-
-## VS Code Preview Extension
-
-{% callout type="info" %}
-We're rebuilding Kilo Code from the ground up on the new [Kilo CLI](https://github.com/Kilo-Org/kilocode). The pre-release extension is available for users who want to try the latest architecture and provide feedback, and don't mind some missing features and rough edges.
-{% /callout %}
-
-The pre-release extension is a complete rebuild featuring:
-
-- A new Solid.js-based UI
-- Deep integration with the Kilo CLI backend
-- Improved session management and model switching
-
-### Current Status
-
-This is an early pre-release. Core features like chat, markdown rendering, authentication, and model/mode switching are working. Some features from the stable extension are still being implemented.
-
-For the full feature status, see the [feature parity tracking document](https://github.com/Kilo-Org/kilocode/blob/main/packages/kilo-vscode/docs/opencode-migration-plan.md).
-
-### Installing the Pre-Release
+The current Kilo Code extension is built on the [Kilo CLI](https://github.com/Kilo-Org/kilocode) and is distributed as the **pre-release version** on the VS Code Marketplace.
1. Open VS Code
2. Go to Extensions (`Ctrl+Shift+X` / `Cmd+Shift+X`)
3. Search for "Kilo Code"
4. Click the dropdown arrow next to **Install** and select **Install Pre-Release Version**
-### Switching Back to Stable
-
-If you need to return to the stable version:
-
-1. Open Extensions in VS Code
-2. Find Kilo Code
-3. Click the dropdown and select **Switch to Release Version**
-
-### Feedback and Issues
-
-Report issues or provide feedback in the [Kilo-Org/kilocode repository](https://github.com/Kilo-Org/kilocode/issues).
-
-{% /tab %}
-{% tab label="JetBrains" %}
-
-## JetBrains IDEs
-
-{% partial file="install-jetbrains.md" /%}
+{% callout type="info" %}
+The "pre-release" label is a VS Code Marketplace distribution channel — the extension is stable and recommended for all users.
+{% /callout %}
{% /tab %}
{% tab label="CLI" %}
@@ -70,6 +32,27 @@ Report issues or provide feedback in the [Kilo-Org/kilocode repository](https://
{% partial file="install-cli.md" /%}
+{% /tab %}
+{% tab label="VS Code (Legacy)" %}
+
+## VS Code Legacy Extension
+
+The legacy extension is the previous version of Kilo Code for VS Code. It is still available but is no longer actively developed. We recommend installing the current extension (see the **VS Code** tab).
+
+To install or switch back to the legacy version:
+
+1. Open VS Code
+2. Go to Extensions (`Ctrl+Shift+X` / `Cmd+Shift+X`)
+3. Search for "Kilo Code"
+4. Click the dropdown arrow next to **Install** and select **Switch to Release Version**
+
+{% /tab %}
+{% tab label="JetBrains" %}
+
+## JetBrains IDEs
+
+{% partial file="install-jetbrains.md" /%}
+
{% /tab %}
{% tab label="Slack" %}
diff --git a/packages/kilo-docs/pages/getting-started/quickstart.md b/packages/kilo-docs/pages/getting-started/quickstart.md
index 3258a77a4d8..5a1a82f2229 100644
--- a/packages/kilo-docs/pages/getting-started/quickstart.md
+++ b/packages/kilo-docs/pages/getting-started/quickstart.md
@@ -5,12 +5,112 @@ description: "Get up and running with Kilo Code in minutes"
# Quickstart
+After you [set up Kilo Code](/docs/getting-started/setup-authentication), follow the guide for your platform below.
+
+{% tabs %}
+{% tab label="VSCode" %}
+
+## Step by Step Guide
+
+### Step 1: Open Kilo Code
+
+Click the Kilo Code icon in the VS Code Primary Side Bar to open the chat panel. If you don't see the icon, verify the [extension is installed](/docs/getting-started/installing).
+
+### Step 2: Type Your Task
+
+Type a clear, concise description of what you want Kilo Code to do in the chat box. The same examples work here:
+
+- "Create a file named `hello.txt` containing 'Hello, world!'."
+- "Write a Python function that adds two numbers."
+- "Create an HTML file for a simple website with the title 'Kilo test'"
+
+No special commands or syntax needed—just use plain English.
+
+### Step 3: Send Your Task
+
+Press **Enter** to send.
+
+### Step 4: Review & Approve Actions
+
+Kilo Code analyzes your request and proposes actions. By default, most tools are auto-approved — only shell commands, external directory access, and sensitive file reads will prompt for confirmation. You'll see the tool name, arguments, and can approve or reject each action.
+
+To change which actions require approval, open **Settings** (gear icon) and go to the **Auto-Approve** tab. You can set each tool to Allow, Ask, or Deny. See [Auto-Approving Actions](/docs/getting-started/settings/auto-approving-actions) for details.
+
+### Step 5: Iterate
+
+Kilo Code works iteratively. Continue giving feedback or follow-up instructions until your task is complete.
+
+### Key Differences from Legacy
+
+- **Settings** are managed via `kilo.jsonc` config files (the Settings webview reads and writes the same files)
+- **Permissions** use a granular per-tool system instead of broad approval categories
+- **Modes** are called "agents" and configured as `.md` files or via the `agent` config key
+- **Autocomplete** uses FIM (Fill-in-the-Middle) with Codestral
+
+{% /tab %}
+{% tab label="CLI" %}
+
+## CLI Quickstart
+
+### Step 1: Open a Terminal
+
+Navigate to your project directory:
+
+```bash
+cd /path/to/your/project
+```
+
+### Step 2: Launch Kilo
+
+Run the `kilo` command to start the interactive TUI (terminal user interface):
+
+```bash
+kilo
+```
+
+If this is your first time, run `kilo auth login` first to authenticate (see [Setup & Authentication](/docs/getting-started/setup-authentication)).
+
+### Step 3: Type Your Task
+
+Type your request in natural language at the prompt. The same examples work here:
+
+- "Create a file named `hello.txt` containing 'Hello, world!'."
+- "Write a Python function that adds two numbers."
+- "Create an HTML file for a simple website with the title 'Kilo test'"
+
+Press **Enter** to send.
+
+### Step 4: Review & Approve Actions
+
+Kilo analyzes your request and proposes actions. By default, most tools are auto-approved — only shell commands, external directory access, and sensitive file reads will prompt for confirmation. You'll see the tool name, arguments, and can approve or reject each action.
+
+To change permission defaults, configure the `permission` key in your `kilo.jsonc` config file. See [Auto-Approving Actions](/docs/getting-started/settings/auto-approving-actions) for details.
+
+### Step 5: Iterate
+
+Kilo works iteratively. Continue giving feedback or follow-up instructions until your task is complete.
+
+### One-Shot Mode
+
+For quick, non-interactive tasks, use `kilo run`:
+
+```bash
+kilo run "add error handling to src/api.ts"
+```
+
+Add `--auto` to auto-approve all permissions (use carefully):
+
+```bash
+kilo run --auto "fix the failing tests in test/auth.test.ts"
+```
+
+{% /tab %}
+{% tab label="VSCode (Legacy)" %}
+
## Video Tour
{% youtube url="https://www.youtube.com/watch?v=pO7zRLQS-p0" caption="This quick tour shows how Kilo Code handles a simple request from start to finish" /%}
-After you [set up Kilo Code](/docs/getting-started/setup-authentication), follow these steps:
-
## Step by Step Guide
### Step 1: Open Kilo Code
@@ -66,6 +166,9 @@ Kilo Code works iteratively. After each action, it waits for your feedback befor
{% image src="/docs/img/your-first-task/your-first-task-8.png" alt="Final result of a completed task showing the iteration process" width="500" caption="After completing the task, Kilo Code shows the final result and awaits your next instruction." /%}
+{% /tab %}
+{% /tabs %}
+
## Conclusion
You've completed your first task. Along the way you learned:
diff --git a/packages/kilo-docs/pages/getting-started/settings/auto-approving-actions.md b/packages/kilo-docs/pages/getting-started/settings/auto-approving-actions.md
index c071327ebc1..10e4a9d0044 100644
--- a/packages/kilo-docs/pages/getting-started/settings/auto-approving-actions.md
+++ b/packages/kilo-docs/pages/getting-started/settings/auto-approving-actions.md
@@ -9,7 +9,243 @@ description: "Configure automatic approval settings for Kilo Code operations"
**Security Warning:** Auto-approve settings bypass confirmation prompts, giving Kilo Code direct access to your system. This can result in data loss, file corruption, or worse. Command line access is particularly dangerous, as it can potentially execute harmful operations that could damage your system or compromise security. Only enable auto-approval for actions you fully trust.
{% /callout %}
-Auto-approve settings speed up your workflow by eliminating repetitive confirmation prompts, but they significantly increase security risks.
+Auto-approve settings speed up your workflow by eliminating repetitive confirmation prompts, but they significantly increase security risks. The **VSCode (Legacy)**, **VSCode**, and **CLI** versions each handle permissions differently — choose the tab that matches your setup.
+
+{% tabs %}
+{% tab label="VSCode" %}
+
+## Overview
+
+The extension uses a granular, per-tool permission system. You can configure permissions through the **Settings → Auto Approve** tab, which provides a UI with per-tool **Allow / Ask / Deny** dropdowns.
+
+The UI reads and writes to the same `kilo.jsonc` config files used by the CLI, so changes made in either place are reflected in both.
+
+## Permission Levels
+
+Each tool permission can be set to one of three values:
+
+| Value | Behavior |
+| --------- | --------------------------------------------------------- |
+| `"allow"` | The tool runs automatically without prompting |
+| `"ask"` | Kilo pauses and asks for approval before running the tool |
+| `"deny"` | The tool is blocked entirely |
+
+When no rule matches a permission check, the default action is `ask`.
+
+## Available Tool Permissions
+
+The Auto Approve tab lists the following tool-specific permissions. Some tools are grouped together in the UI and share a single permission level:
+
+| Permission | Controls |
+| -------------------------- | ------------------------------------------------------ |
+| `external_directory` | Accessing files outside the project directory |
+| `bash` | Executing shell commands |
+| `read` | Reading file contents |
+| `edit` | Editing existing files |
+| `glob` | File pattern matching / searching by name |
+| `grep` | Searching file contents by regex |
+| `list` | Listing directory contents |
+| `task` | Launching sub-agents |
+| `skill` | Loading specialized skills |
+| `lsp` | Language server protocol operations |
+| `todoread` / `todowrite` | Reading and updating the todo list |
+| `websearch` / `codesearch` | Performing web or code searches |
+| `webfetch` | Fetching content from URLs |
+| `doom_loop` | Allowing the agent to continue after repeated failures |
+
+## Runtime Permission Requests
+
+When a tool is set to `"ask"`, Kilo pauses and displays a permission prompt with two options:
+
+| Option | Behavior |
+| -------- | ------------------------------ |
+| **Run** | Allow this specific invocation |
+| **Deny** | Block this specific invocation |
+
+Expand **Manage Auto-Approve Rules** to add commands or patterns to your allowed or denied lists. These rules are then appended to the bottom of the approval rules in settings and the config file.
+
+## Defaults
+
+Most tools default to `"*": "allow"` for a smooth out-of-the-box experience. Notable exceptions that prompt by default:
+
+- **`.env` files** — reading `.env` files prompts for approval. Files matching `*.env.*` (e.g., `.env.local`, `.env.production`) also trigger an ask, while `*.env.example` is explicitly allowed.
+- **`external_directory`** — accessing files outside the project prompts for approval
+- **`doom_loop`** — prompts when the agent enters a repeated failure cycle
+
+{% /tab %}
+{% tab label="CLI" %}
+
+## Overview
+
+The CLI uses a granular, per-tool permission system configured in `kilo.jsonc`. Instead of broad categories like "read" or "write," each tool has its own permission level with glob-pattern rules for fine-grained control.
+
+## Permission Levels
+
+Each tool permission can be set to one of three values:
+
+| Value | Behavior |
+| --------- | --------------------------------------------------------- |
+| `"allow"` | The tool runs automatically without prompting |
+| `"ask"` | Kilo pauses and asks for approval before running the tool |
+| `"deny"` | The tool is blocked entirely |
+
+When no rule matches a permission check, the default action is `ask`.
+
+## Available Tool Permissions
+
+Permissions are configured under the `permission` key in `kilo.jsonc`. The following tool-specific permission levels are available:
+
+| Permission | Controls |
+| -------------------------- | ------------------------------------------------------ |
+| `external_directory` | Accessing files outside the project directory |
+| `bash` | Executing shell commands |
+| `read` | Reading file contents |
+| `edit` | Editing existing files |
+| `glob` | File pattern matching / searching by name |
+| `grep` | Searching file contents by regex |
+| `list` | Listing directory contents |
+| `task` | Launching sub-agents |
+| `skill` | Loading specialized skills |
+| `lsp` | Language server protocol operations |
+| `todoread` / `todowrite` | Reading and updating the todo list |
+| `websearch` / `codesearch` | Performing web or code searches |
+| `webfetch` | Fetching content from URLs |
+| `doom_loop` | Allowing the agent to continue after repeated failures |
+
+## Glob-Pattern Rules
+
+Instead of a simple `"allow"` or `"deny"`, each tool can use glob-pattern rules for granular control. Patterns are matched against the tool's arguments (command strings, file paths, etc.), and the last matching rule wins.
+
+### Example: Shell Commands
+
+Allow git commands automatically, but prompt for everything else:
+
+```json
+{
+ "permission": {
+ "bash": {
+ "git *": "allow",
+ "*": "ask"
+ }
+ }
+}
+```
+
+### Example: File Reading
+
+Prompt before reading `.env` files, but allow all other reads:
+
+```json
+{
+ "permission": {
+ "read": {
+ "*.env": "ask",
+ "*": "allow"
+ }
+ }
+}
+```
+
+### Example: Blocking Dangerous Commands
+
+Deny `rm -rf` commands, allow common dev commands, and ask for anything else:
+
+```json
+{
+ "permission": {
+ "bash": {
+ "rm -rf *": "deny",
+ "npm *": "allow",
+ "bun *": "allow",
+ "git *": "allow",
+ "*": "ask"
+ }
+ }
+}
+```
+
+## Per-Agent Permission Overrides
+
+Different agents can have different permission levels. Override the default permissions for a specific agent under the `agent.
+
+ )
+ }
const [title, ...rest] = cleaned.split(": ")
return (
+
-
+
}
>
-
+
)}
diff --git a/packages/kilo-ui/src/components/text-shimmer.tsx b/packages/kilo-ui/src/components/text-shimmer.tsx
index 8a835e07cb7..382d1ec27d6 100644
--- a/packages/kilo-ui/src/components/text-shimmer.tsx
+++ b/packages/kilo-ui/src/components/text-shimmer.tsx
@@ -1 +1 @@
-export * from "@opencode-ai/ui/text-shimmer"
\ No newline at end of file
+export * from "@opencode-ai/ui/text-shimmer"
diff --git a/packages/kilo-ui/src/components/tool-count-summary.tsx b/packages/kilo-ui/src/components/tool-count-summary.tsx
index 63aed1c7710..52bc991829c 100644
--- a/packages/kilo-ui/src/components/tool-count-summary.tsx
+++ b/packages/kilo-ui/src/components/tool-count-summary.tsx
@@ -1 +1 @@
-export * from "@opencode-ai/ui/tool-count-summary"
\ No newline at end of file
+export * from "@opencode-ai/ui/tool-count-summary"
diff --git a/packages/kilo-ui/src/hooks/create-auto-scroll.tsx b/packages/kilo-ui/src/hooks/create-auto-scroll.tsx
index 99862e4ae73..6ea1f5d8ad0 100644
--- a/packages/kilo-ui/src/hooks/create-auto-scroll.tsx
+++ b/packages/kilo-ui/src/hooks/create-auto-scroll.tsx
@@ -7,7 +7,6 @@ const DEBOUNCE_MS = 100
export interface AutoScrollOptions {
working: () => boolean
onUserInteracted?: () => void
- overflowAnchor?: "none" | "auto" | "dynamic"
bottomThreshold?: number
}
@@ -15,10 +14,10 @@ export function createAutoScroll(options: AutoScrollOptions) {
let scroll: HTMLElement | undefined
let settling = false
let settleTimer: ReturnType | undefined
- let autoTimer: ReturnType | undefined
let stopTimer: ReturnType | undefined
let cleanup: (() => void) | undefined
- let auto: { time: number } | undefined
+ let userInitiated = false
+ let lastScrollTop: number | undefined
const threshold = () => options.bottomThreshold ?? 10
@@ -37,36 +36,18 @@ export function createAutoScroll(options: AutoScrollOptions) {
return el.scrollHeight - el.clientHeight > 1
}
- // Browsers can dispatch scroll events asynchronously. If new content arrives
- // between us calling `scrollTo()` and the subsequent `scroll` event firing,
- // the handler can see a non-zero `distanceFromBottom` and incorrectly assume
- // the user scrolled.
- const markAuto = (_el: HTMLElement) => {
- auto = { time: Date.now() }
-
- if (autoTimer) clearTimeout(autoTimer)
- autoTimer = setTimeout(() => {
- auto = undefined
- autoTimer = undefined
- }, 250)
- }
-
- const isAuto = (_el: HTMLElement) => {
- const a = auto
- if (!a) return false
-
- if (Date.now() - a.time > 250) {
- auto = undefined
- return false
+ const markUser = (e: Event) => {
+ if (e instanceof WheelEvent) {
+ const target = e.target instanceof Element ? e.target : undefined
+ const nested = target?.closest("[data-scrollable]")
+ if (scroll && nested && nested !== scroll) return
}
-
- return true
+ userInitiated = true
}
const scrollToBottomNow = (behavior: ScrollBehavior) => {
const el = scroll
if (!el) return
- markAuto(el)
if (behavior === "smooth") {
el.scrollTo({ top: el.scrollHeight, behavior })
return
@@ -74,6 +55,7 @@ export function createAutoScroll(options: AutoScrollOptions) {
// `scrollTop` assignment bypasses any CSS `scroll-behavior: smooth`.
el.scrollTop = el.scrollHeight
+ lastScrollTop = el.scrollTop
}
const scrollToBottom = (force: boolean) => {
@@ -121,19 +103,28 @@ export function createAutoScroll(options: AutoScrollOptions) {
const el = scroll
if (!el) return
+ const byUser = userInitiated
+ userInitiated = false
+ const distance = distanceFromBottom(el)
+
if (!canScroll(el)) {
if (store.userScrolled) setStore("userScrolled", false)
return
}
- if (distanceFromBottom(el) < threshold()) {
+ if (distance < threshold()) {
if (store.userScrolled) setStore("userScrolled", false)
+ lastScrollTop = el.scrollTop
return
}
- // Ignore scroll events triggered by our own scrollToBottom calls.
- if (!store.userScrolled && isAuto(el)) {
- scrollToBottom(false)
+ if (!store.userScrolled && !byUser) {
+ if (el.scrollTop < (lastScrollTop ?? el.scrollTop)) {
+ stop()
+ } else {
+ scrollToBottomNow("auto")
+ }
+ lastScrollTop = el.scrollTop
return
}
@@ -145,7 +136,6 @@ export function createAutoScroll(options: AutoScrollOptions) {
const cur = scroll
if (!cur) return
if (distanceFromBottom(cur) < threshold()) return
- if (!store.userScrolled && isAuto(cur)) return
stop()
}, DEBOUNCE_MS)
}
@@ -155,22 +145,6 @@ export function createAutoScroll(options: AutoScrollOptions) {
stop()
}
- const updateOverflowAnchor = (el: HTMLElement) => {
- const mode = options.overflowAnchor ?? "dynamic"
-
- if (mode === "none") {
- el.style.overflowAnchor = "none"
- return
- }
-
- if (mode === "auto") {
- el.style.overflowAnchor = "auto"
- return
- }
-
- el.style.overflowAnchor = store.userScrolled ? "auto" : "none"
- }
-
createResizeObserver(
() => store.contentRef,
() => {
@@ -179,8 +153,16 @@ export function createAutoScroll(options: AutoScrollOptions) {
if (store.userScrolled) setStore("userScrolled", false)
return
}
- if (!active()) return
- if (store.userScrolled) return
+ if (!active()) {
+ if (!store.userScrolled && el && distanceFromBottom(el) > threshold()) {
+ scrollToBottomNow("auto")
+ return
+ }
+ return
+ }
+ if (store.userScrolled) {
+ return
+ }
// ResizeObserver fires after layout, before paint.
// Keep the bottom locked in the same frame to avoid visible
// "jump up then catch up" artifacts while streaming content.
@@ -206,18 +188,8 @@ export function createAutoScroll(options: AutoScrollOptions) {
}),
)
- createEffect(() => {
- // Track `userScrolled` even before `scrollRef` is attached, so we can
- // update overflow anchoring once the element exists.
- store.userScrolled
- const el = scroll
- if (!el) return
- updateOverflowAnchor(el)
- })
-
onCleanup(() => {
if (settleTimer) clearTimeout(settleTimer)
- if (autoTimer) clearTimeout(autoTimer)
if (stopTimer) clearTimeout(stopTimer)
if (cleanup) cleanup()
})
@@ -233,11 +205,19 @@ export function createAutoScroll(options: AutoScrollOptions) {
if (!el) return
- updateOverflowAnchor(el)
+ el.style.overflowAnchor = "auto"
el.addEventListener("wheel", handleWheel, { passive: true })
+ el.addEventListener("wheel", markUser, { passive: true, capture: true })
+ el.addEventListener("pointerdown", markUser, { passive: true })
+ el.addEventListener("keydown", markUser, { passive: true })
+ el.addEventListener("touchstart", markUser, { passive: true })
cleanup = () => {
el.removeEventListener("wheel", handleWheel)
+ el.removeEventListener("wheel", markUser, { capture: true })
+ el.removeEventListener("pointerdown", markUser)
+ el.removeEventListener("keydown", markUser)
+ el.removeEventListener("touchstart", markUser)
}
},
contentRef: (el: HTMLElement | undefined) => setStore("contentRef", el),
diff --git a/packages/kilo-ui/src/stories/basic-tool.stories.tsx b/packages/kilo-ui/src/stories/basic-tool.stories.tsx
index ac8252e077e..255ea5fe13a 100644
--- a/packages/kilo-ui/src/stories/basic-tool.stories.tsx
+++ b/packages/kilo-ui/src/stories/basic-tool.stories.tsx
@@ -40,12 +40,7 @@ export const NoChildren: Story = {
export const Locked: Story = {
render: () => (
-
+
),
diff --git a/packages/kilo-ui/src/stories/message-part.stories.tsx b/packages/kilo-ui/src/stories/message-part.stories.tsx
index 112e29e236a..e67fe32b441 100644
--- a/packages/kilo-ui/src/stories/message-part.stories.tsx
+++ b/packages/kilo-ui/src/stories/message-part.stories.tsx
@@ -353,3 +353,91 @@ export const WithContextGroup: Story = {
),
}
+
+// --- All 5 tool hint error types in a single screenshot ---
+
+const hintErrors: ToolPart[] = [
+ {
+ id: "part-hint-001",
+ sessionID: SESSION_ID,
+ messageID: ASST_MSG_ID,
+ type: "tool",
+ callID: "call-hint-001",
+ tool: "write",
+ state: {
+ status: "error",
+ input: { filePath: "src/config.ts" },
+ error:
+ "Error: File already exists — read it before overwriting it. Use the Read tool first to see its current contents.",
+ time: { start: now - 6000, end: now - 5500 },
+ },
+ },
+ {
+ id: "part-hint-002",
+ sessionID: SESSION_ID,
+ messageID: ASST_MSG_ID,
+ type: "tool",
+ callID: "call-hint-002",
+ tool: "edit",
+ state: {
+ status: "error",
+ input: { filePath: "src/app.tsx" },
+ error: "Error: File has been modified since it was last read. Please read it again before editing.",
+ time: { start: now - 5500, end: now - 5000 },
+ },
+ },
+ {
+ id: "part-hint-003",
+ sessionID: SESSION_ID,
+ messageID: ASST_MSG_ID,
+ type: "tool",
+ callID: "call-hint-003",
+ tool: "edit",
+ state: {
+ status: "error",
+ input: { filePath: "src/utils.ts" },
+ error: "Error: oldString and newString are identical. No changes were made.",
+ time: { start: now - 5000, end: now - 4500 },
+ },
+ },
+ {
+ id: "part-hint-004",
+ sessionID: SESSION_ID,
+ messageID: ASST_MSG_ID,
+ type: "tool",
+ callID: "call-hint-004",
+ tool: "edit",
+ state: {
+ status: "error",
+ input: { filePath: "src/index.ts" },
+ error:
+ "Error: oldString not found in file. The oldString must match exactly, including whitespace and indentation.",
+ time: { start: now - 4500, end: now - 4000 },
+ },
+ },
+ {
+ id: "part-hint-005",
+ sessionID: SESSION_ID,
+ messageID: ASST_MSG_ID,
+ type: "tool",
+ callID: "call-hint-005",
+ tool: "edit",
+ state: {
+ status: "error",
+ input: { filePath: "src/counter.tsx" },
+ error:
+ "Error: Found multiple matches for oldString. Provide more surrounding lines to identify the correct match.",
+ time: { start: now - 4000, end: now - 3500 },
+ },
+ },
+]
+
+const mockDataHintErrors = createMockData(hintErrors)
+
+export const ToolHintErrors: Story = {
+ render: () => (
+
+
+ ),
+}
diff --git a/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-messagepart/full-conversation-turn-chromium-linux.png b/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-messagepart/full-conversation-turn-chromium-linux.png
index 6f383cb9019..27646a659da 100644
--- a/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-messagepart/full-conversation-turn-chromium-linux.png
+++ b/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-messagepart/full-conversation-turn-chromium-linux.png
@@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1
-oid sha256:9cad4954b9ac5c0d1a245d14b3ab51836217479870c39589a55eb862c4247aa9
-size 18902
+oid sha256:81fd232992e804e8e0b3453c4e77d358f8c24dd7495b354b8d7de24c7a43724c
+size 18986
diff --git a/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-messagepart/tool-hint-errors-chromium-linux.png b/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-messagepart/tool-hint-errors-chromium-linux.png
new file mode 100644
index 00000000000..4650c65cafe
--- /dev/null
+++ b/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-messagepart/tool-hint-errors-chromium-linux.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:958b3b690c8e19c8c4ffc451da476954e9334f5619a1b8b40ce30da848240f6d
+size 15354
diff --git a/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-messagepart/user-message-story-chromium-linux.png b/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-messagepart/user-message-story-chromium-linux.png
index b68a8dc340c..c11138b36db 100644
--- a/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-messagepart/user-message-story-chromium-linux.png
+++ b/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-messagepart/user-message-story-chromium-linux.png
@@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1
-oid sha256:876fba84b341bffa55bedb3c6d3384c5099c9a1d9336a3090b429b87e1256110
-size 4149
+oid sha256:4b690e1cba9e1a7a4eef86162c762092f5908f2d498f0d77f48c44fef1fd44ff
+size 4242
diff --git a/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-sessionturn/default-chromium-linux.png b/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-sessionturn/default-chromium-linux.png
index 4b660e94928..5e3dcecb255 100644
--- a/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-sessionturn/default-chromium-linux.png
+++ b/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-sessionturn/default-chromium-linux.png
@@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1
-oid sha256:0fc349b25120ba40eaf3f6e6cb9adf268a0433c6a4b4e723f5a9957c555aec25
-size 29267
+oid sha256:71a4c5ce0d2bd00ac8428a65e3c4db6c0455720cac5467f8f483a9708cea8765
+size 29370
diff --git a/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-sessionturn/with-steps-expanded-chromium-linux.png b/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-sessionturn/with-steps-expanded-chromium-linux.png
index 7adc3a5b82c..5e3dcecb255 100644
--- a/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-sessionturn/with-steps-expanded-chromium-linux.png
+++ b/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-sessionturn/with-steps-expanded-chromium-linux.png
@@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1
-oid sha256:c67bd29525bc38133cc5271d0467e85e131a0bd41313a4033836e7a6bc5e02e2
-size 26312
+oid sha256:71a4c5ce0d2bd00ac8428a65e3c4db6c0455720cac5467f8f483a9708cea8765
+size 29370
diff --git a/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-sessionturn/working-chromium-linux.png b/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-sessionturn/working-chromium-linux.png
index a69883f2758..6d44ffd8a25 100644
--- a/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-sessionturn/working-chromium-linux.png
+++ b/packages/kilo-ui/tests/visual-regression.spec.ts-snapshots/components-sessionturn/working-chromium-linux.png
@@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1
-oid sha256:640fabcec21e9e8e2251e1b347966790d347af9c0b438ca4dc9bc118819069f2
-size 4562
+oid sha256:4272e7e538d95ee3c7010dca5711a53643573c0efed2c527d6db58ef23d6c086
+size 4627
diff --git a/packages/kilo-vscode/AGENTS.md b/packages/kilo-vscode/AGENTS.md
index 3604d6c14b4..ff630a3ac06 100644
--- a/packages/kilo-vscode/AGENTS.md
+++ b/packages/kilo-vscode/AGENTS.md
@@ -201,6 +201,16 @@ New webview features must use **`@kilocode/kilo-ui`** components instead of raw
This package is entirely Kilo-specific — `kilocode_change` markers are NOT needed in any files under `packages/kilo-vscode/`. The markers are only necessary when modifying shared upstream opencode files.
+## Process Spawning (Windows)
+
+On Windows, any `spawn`/`execFile`/`exec` call that does not set `windowsHide: true` will flash a cmd.exe console window at the user. To prevent this, **never import `spawn`, `execFile`, or `exec` from `child_process` directly**. Use the wrappers in `src/util/process.ts` instead — they enforce `windowsHide: true` automatically:
+
+```ts
+import { spawn, exec } from "../util/process"
+```
+
+The `spawn` wrapper covers long-lived processes (e.g. `kilo serve`). The `exec` wrapper covers short commands (e.g. `git`, `tar`). If you need the raw callback form of `execFile` for some reason, pass `windowsHide: true` explicitly in the options object.
+
## Style
Follow monorepo root AGENTS.md style guide:
diff --git a/packages/kilo-vscode/LICENSE b/packages/kilo-vscode/LICENSE
new file mode 100644
index 00000000000..5cb50685f18
--- /dev/null
+++ b/packages/kilo-vscode/LICENSE
@@ -0,0 +1,22 @@
+MIT License
+
+Copyright (c) 2025 Kilo Code
+Copyright (c) 2025 opencode
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
diff --git a/packages/kilo-vscode/README.md b/packages/kilo-vscode/README.md
index aef1c95a4d2..e69a35887f8 100644
--- a/packages/kilo-vscode/README.md
+++ b/packages/kilo-vscode/README.md
@@ -1,9 +1,9 @@
-
+
+
+ {"Hello, world!"}
- 
+
+
+
+