## Problem
Anthropic returns HTTP 400 when an assistant message contains a
`web_search_tool_result` block whose `tool_use_id` has no matching
earlier `server_tool_use` block in the same assistant message. A
previous fix (#24706) sanitized provider-executed tool calls without
matching results, but the opposite direction, orphaned or misordered
provider-executed results, could still slip through both the prompt
sanitizer and the persistence path.
## Fix
Tighten Anthropic provider-executed tool history handling while
preserving the useful result payload as normal assistant text when the
provider-tool metadata is unsafe.
1. Extract Anthropic provider-tool sanitization into
`coderd/x/chatd/chatsanitize` so provider-specific repair logic is no
longer spread through `chatprompt` and `chatloop`.
2. `chatsanitize.SanitizeAnthropicProviderToolHistory` removes invalid
provider-executed tool structure for Anthropic prompts: orphans in
either direction, result-before-call, duplicate IDs, invalid JSON
inputs, empty IDs and tool names, unsupported tool names, mismatched
`ProviderExecuted` flags, provider-executed blocks outside assistant
messages, and web-search results without serializable Anthropic result
metadata. Provider-executed result payloads are textified instead of
being discarded when there is text to preserve.
3. `chatsanitize.SanitizeAnthropicProviderToolContent` mirrors the same
rule at the streamed step content level. Persisted history no longer
carries invalid provider-tool blocks forward, but it keeps the result
text for future turns.
4. `chatsanitize.ApplyAnthropicProviderToolGuard` only repairs
structurally invalid Anthropic provider-tool history. It no longer
strips otherwise-valid historical `web_search` blocks just because web
search is disabled for the current request. The fail-closed fallback
also textifies provider results before removing provider-tool metadata.
Tests cover prompt sanitization, validation reason strings, result
payload textification, content-level persistence sanitization, disabled
web-search history preservation, direct pre-request guard behavior, and
the fallback strip path.
> Mux is acting on Mike's behalf.
Persists the agents page archived filter in the URL via
`?archived=archived`,
so deep-linking to archived agents and restoring the filter from history
work
as expected. Unknown values fall back to active. Toggling back to active
removes the param from the URL so the default state has one canonical
form.
Also fixes a regression that surfaced once the filter became
URL-derived:
clicking a chat in the sidebar previously navigated to `/agents/:id`
with no
search params, silently resetting the filter. The sidebar's chat
`NavLink`
now preserves `location.search`.
Coverage:
- `useArchivedFilterParam` is unit-tested with `renderHook`, covering
URL
parsing variants and the `deleteValue` semantics for the default state.
- Cross-route preservation is covered by a Storybook play story
(`PreservesArchivedFilterOnChatNavigation`) that renders the real
`NavLink`
and asserts on a probe child route.
- The pre-existing sidebar callback tests in `AgentsSidebar.test.tsx`
cover
wiring that exists on `main`; they're kept here for completeness rather
than as new feature gates.
_Generated by Coder Agent._
Follow-up to #24650.
Canceling the terminal command confirmation dialog now calls
`window.close()` instead of stripping the `?command=` query parameter
and opening a plain terminal. The terminal always opens in a new tab,
so closing it is the expected UX when the user declines.
> 🤖 Generated by Coder Agents
Depends on #24642
Adds per-owner digest notifications onto the chat auto-archive
subsystem.
Each tick's archived rows are grouped by owner, the top 25 titles per
owner are rendered into a new `Chats Auto-Archived` notification
template, and any remainder surfaces as `and N more`. Each digest is
per-tick, so users with large amounts of purgeable data may get multiple
notifications in sequence (one per user per tick).
The template body branches on `retention_days`: when retention is
disabled (`retention_days=0`), users are told archived chats are kept
indefinitely rather than falsely claiming imminent deletion.
### Changes
- migration `000XXX_chat_auto_archive_notification_template` adds new
notification template
- `dbpurge`: threads `notifications.Enqueuer` through `New`; and
enqueues notification message.
- `cli/server.go`: passes `options.NotificationsEnqueuer` into
`dbpurge.New`.
- `coderd/notifications/events.go`: new `TemplateChatAutoArchiveDigest`
UUID.
- `coderd/inboxnotifications.go`: inbox registration.
- Docs: adds a `Notifications` section to `chat-auto-archive.md`.
> 🤖
> Mux working on behalf of Mike.
## Summary
Clarifies that GitHub PR body prose should rely on soft wrapping instead
of manual fixed-width hard wraps. Updates the examples and key
principles to match that guidance.
## Validation
- `pnpm exec markdownlint-cli2 .claude/docs/PR_STYLE_GUIDE.md`
- `make lint/emdash`
- `make pre-commit-light`
The `--login-type none` option for `coder users create` is deprecated.
This adds deprecation warnings to all docs that reference it and updates
the CI/CD tutorial to recommend the replacement flows.
Refs DEVEX-224
<details>
<summary>Changes</summary>
- `cli/usercreate.go`: Append deprecation notice to `--login-type` flag
description.
- `docs/tutorials/testing-templates.md`: Replace `--login-type none`
example with separate Premium (`--service-account`) and OSS
(`--login-type password`) examples.
- `docs/reference/cli/users_create.md`: Regenerated from CLI source.
- `cli/testdata/coder_users_create_--help.golden`: Updated golden
snapshot.
</details>
> [!NOTE]
> Generated by Coder Agents.
Currently when a user clicks either the Cancel or Allow button on the
authorization page the client app URI is executed but the page does not
land to the main dashboard page, leaving the two buttons open for
multiple clicks from the user. Aside from the potential problems it
might cause by activating the callback URI multiple times, the page also
provides poor UX because users usually expect the authorization tab to
return to the dashboard.
The consent page now executes the OAuth2 callback (auth code on Allow,
`access_denied` on Cancel) and hides the two buttons and updates the
existing description with a user instruction to close the window.
Initial implementation relied on a pop-up window executing the callback
while the main window was redirected to the dashboard main page.
- resolves https://github.com/coder/coder/issues/20323
<!--
If you have used AI to produce some or all of this PR, please ensure you
have read our [AI Contribution
guidelines](https://coder.com/docs/about/contributing/AI_CONTRIBUTING)
before submitting.
-->
The agent chat thinking disclosure used a smaller label with the caret
on the left, which made collapsed and in-progress thinking look
different from collapsible tool calls.
Align the thinking disclosure with the shared tool-call row treatment by
using the same label size, trailing caret placement, and hover color
while preserving the streaming shimmer. Adds a Storybook story that
renders collapsed thinking next to a tool call.
<details>
<summary>Storybook screenshots</summary>
Captured from Storybook:
-
`pages/AgentsPage/ChatConversation/ConversationTimeline/ThinkingBlockWithToolCall`
-
`pages/AgentsPage/ChatConversation/ConversationTimeline/ThinkingBlockWithToolCall`
hovered
-
`pages/AgentsPage/ChatConversation/StreamingOutput/ThinkingDuringStreamingWithToolCalls`
Screenshots are attached in the Coder task.
</details>
Generated by Coder Agents.
The deprecation notice on the [MCP Tools
Injection](https://coder.com/docs/ai-coder/ai-gateway/mcp) page
currently states the feature "will be removed in a future release,"
which may cause concern for users relying on it today.
This updates the warning to clarify that the feature will remain
functional and will not be removed until its replacement, MCP Gateway,
is released.
> [!NOTE]
> Generated by Coder Agents
---------
Co-authored-by: david-fraley <67079030+david-fraley@users.noreply.github.com>
> Mux is acting on Mike's behalf.
Adds explicit star, edit, and duplicate actions to each Agents model
configuration row, replacing the chevron-only affordance.
Duplicate opens a prefilled create form backed by the existing create
mutation when the provider can manage models. The form copies editable
model fields and provider config while clearing default status so saving
a duplicate does not change the current default model.
Bumps coder/terraform-provider-coder to v2.16.0 to pick up the `coder_secret`
data source that enables expressing a required user secret in a template.
This change passes user secrets from coderd to the Terraform process at
workspace build time so the `data.coder_secret` data source in
terraform-provider-coder can resolve values at plan time.
Secrets traverse two proto hops: `provisionerdserver` fetches them
via`ListUserSecretsWithValues`, attaches them to
`AcquiredJob.WorkspaceBuild.user_secrets` on `provisionerd.proto`;
`runner.go` forwards into `PlanRequest.user_secrets` on
`provisioner.proto`; the Terraform provisioner encodes each as
`CODER_SECRET_ENV_<name>` or `CODER_SECRET_FILE_<hex(path)>` before
invoking `terraform plan`. Only plan requests carry secrets; apply runs
with `nil` because values are baked into plan state.
Fetch is gated on a workspace transitioning to start. stop and delete
transitions never carry secrets, so revoking or deleting a stored secret
cannot make a workspace unstoppable. DB errors on the fetch fail the job
outright rather than silently continuing with an empty secret set.
Note that user secrets will be stored in the workspace_builds table in
provisioner_state with other Terraform state (including other sensitive data).
The terminal page auto-executed commands from the `?command=` query
parameter
on page load without user confirmation. Because session auth uses
`SameSite=Lax`
cookies, an attacker could craft a link (phishing email, Slack DM,
external page)
that executes arbitrary commands in a victim's workspace when clicked.
Adds a `ConfirmDialog` that shows the exact command and requires
explicit user
approval before it is passed to the terminal WebSocket. Canceling
removes the
`command` parameter from the URL and opens a plain terminal.
<details>
<summary>Implementation details</summary>
### Data flow (before)
`TerminalPage.tsx` reads `searchParams.get("command")` and passes it
directly
as `initialCommand` to `WorkspaceTerminal`, which embeds it in the
WebSocket
URL. `proxy.go` forwards it to the agent, which runs `bash -c
"<command>"`
immediately.
### Fix
- Added `commandConfirmed` state and `commandPendingConfirmation` flag
in
`TerminalPage.tsx`.
- The `loading` prop passed to `WorkspaceTerminal` includes
`commandPendingConfirmation`, keeping the terminal in loading state
until
the user confirms or cancels.
- The command is only passed as `initialCommand` after the user clicks
"Run command" in the confirmation dialog.
- Trusted `?app=` commands (resolved from agent apps) bypass the dialog.
- Cancel removes the `?command=` parameter from the URL entirely.
- No backend changes needed; the frontend gates the command before it
reaches the WebSocket.
### Terminal focus after dialog
`WorkspaceTerminal`'s autoFocus effect previously depended on
`[terminal, isVisible, autoFocus]` but not `loading`. It fired while the
Radix dialog's focus trap was active, so `terminal.focus()` was
intercepted. When `loading` became false after confirming the dialog,
the
effect did not re-fire. Fixed by adding `loading` to the effect deps and
skipping focus while `loading` is true.
### Files changed
| File | Change |
|------|--------|
| `site/src/pages/TerminalPage/TerminalPage.tsx` | Confirmation dialog,
`commandPendingConfirmation` in loading prop |
| `site/src/pages/TerminalPage/TerminalCommandConsentDialog.tsx` | New
dialog component |
| `site/src/pages/TerminalPage/TerminalCommandConsentDialog.stories.tsx`
| Storybook story for dialog |
| `site/src/pages/TerminalPage/TerminalPage.stories.tsx` |
`CommandConfirmation` story |
| `site/src/pages/TerminalPage/TerminalPage.test.tsx` | 4 new dialog
tests, `renderTerminalRaw` helper for non-blocking render |
| `site/src/modules/terminal/WorkspaceTerminal.tsx` | Add `loading` to
autoFocus effect deps |
| `site/e2e/helpers.ts` | Dismiss dialog in `openTerminalWindow` helper
|
| `site/e2e/tests/webTerminal.spec.ts` | Wait for
`data-status="connected"` + click terminal for focus |
</details>
> 🤖 Generated by Coder Agents
---------
Co-authored-by: Jakub Domeracki <jakub@coder.com>
`namedWorkspace` in `cli/root.go` parsed workspace identifiers with
`uuid.Parse` first and returned immediately on success, even when no
workspace had that UUID as its actual ID. This caused 404 errors for any
workspace whose name was a valid 32-char hex string (dashless UUID).
- Add `codersdk.ResolveWorkspace`: tries UUID lookup first, falls back
to name lookup on 404. `NameValid` guard skips the fallback for standard
dashed UUIDs (36 chars > 32-char name limit).
- Export `codersdk.SplitWorkspaceIdentifier`, replacing the duplicate
`splitNamedWorkspace` in `cli/root.go` (uses `strings.Cut`).
- Delete `namedWorkspace` from `cli/root.go`; all 28 call sites now use
`client.ResolveWorkspace` directly.
- Delete `namedWorkspace` and `splitNameAndOwner` from
`codersdk/toolsdk/bash.go`; inline `client.ResolveWorkspace`.
- Simplify `GetWorkspace` tool handler to a single `ResolveWorkspace`
call.
- Unit tests via httptest mock cover UUID, name, owner/name, UUID-like
fallback, not-found, server error, transport error, and invalid
identifier paths.
- Integration tests in `cli/show_test.go` and `codersdk/toolsdk` for
workspaces with UUID-like names.
> Generated with Coder Agents
closes CODAGT-142
The Agents right-panel tabs (Git, Terminal, Desktop) reset to the
default tab (Git) every time the user switches between agent sessions.
This happens because `KeyedAgentChatPage` forces a full remount on
`agentId` change, discarding the `useState` that holds the active tab.
Persist the active tab per agent session in `localStorage`, following
the existing `agents.draft-input.<chatID>` pattern. When the user
returns to a session, the last tab they selected is restored. If the
stored tab is temporarily unavailable (e.g. Desktop while the workspace
is stopped), `SidebarTabView`'s existing fallback to the first
available tab applies, and the stored value survives so it can be
honoured once that tab reappears.
Archiving a chat clears its stored tab entry so unarchiving starts
fresh with the default tab.
Relates to #24642
Adds admin UI controls for managing chat auto-archive (days) under
"Lifecycle".
Also adds a "Days" label to the right of the pre-existing unitless
numeric input for consistency.
Exemplary screenshot below. More screens available in Storybook.
<img width="847" height="585" alt="Screenshot 2026-04-24 at 16 48 59"
src="https://github.com/user-attachments/assets/d38de5f8-d379-4b06-b175-ac399f31e578"
/>
Bedrock chat provider requests can inherit Anthropic public API headers
from the process environment, which causes mixed Anthropic and Bedrock
auth headers on signed requests.
Update the Anthropic SDK fork so its Bedrock middleware strips
Anthropic-only headers before signing requests, and keep a chatprovider
regression test for the production request shape.
> Mux is acting on Mike's behalf.
> Worked on by Mux on Mike's behalf.
## Summary
- Disable OpenAI Responses `previous_response_id` chain mode when the
prior assistant response has unresolved local tool calls, so the next
request can include paired tool outputs instead of sending an incomplete
continuation.
- Update the fantasy pin to a Responses replay fix that preserves stored
reasoning references, only replays web search references when paired
with reasoning, and validates local function-call output pairing before
send.
- Add fake OpenAI Responses input validation for the two production 400
shapes and integration coverage for full-history reasoning plus web
search replay.
- Add sanitized diagnostics for the OpenAI Responses continuity errors.
## Tests
- `go test ./providers/openai -run
'TestResponsesToPrompt_(ReasoningWithStore|ReasoningWithWebSearchCombined|WebSearchRequiresReasoningReference|ReasoningWithFunctionCallCombined|WebSearchProviderExecutedToolResults)|TestPrepareParams_(SkipsProviderExecutedToolReferences|ValidatesFunctionCallOutputPairing)|TestValidateResponsesInput_WebSearchReferenceRequiresReasoning'
-count=1`
- `go test ./providers/openai -count=1`
- `GOWORK=off go test ./coderd/x/chatd/chattest -run
TestValidateResponsesAPIInput -count=1`
- `GOWORK=off go test ./coderd/x/chatd -run
'TestOpenAIResponses(NoStaleWebSearchReplay|FullReplayPairsReasoningAndWebSearch|ChainModeSkipsWhenLocalCallPending|ChainModeStillFiresForProviderExecutedOnly)$|TestResolveChainMode_'
-count=1`
- `GOWORK=off go test ./coderd/x/chatd/chatprompt -run
'TestInjectMissingToolResults_' -count=1`
- `GOWORK=off go test ./coderd/x/chatd/chaterror -run
TestClassify_OpenAIResponsesAPIDiagnostics -count=1`
- `GOWORK=off go test ./coderd/x/chatd/... -count=1`
- `git diff --check`
- `git commit` pre-commit hook
`tryAutoPromoteQueuedMessage` in `processChat`'s deferred cleanup could
set a chat back to `pending` without waking the processor. The processor
only noticed on the next 10ms poll, so under load tests like
`TestAutoPromoteQueuedMessageFallsBackForInvalidQueuedModelConfigID`
could time out waiting for the second streaming request (#1500).
Call `p.signalWake()` after the promoted-message publishes when
`promotedMessage != nil`, matching the pattern used by `CreateChat`,
`SendMessage`, `EditMessage`, `PromoteQueued`, and `InterruptChat`. Make
the regression helper `testAutoPromoteQueuedMessageFallback`
deterministic by setting `PendingChatAcquireInterval = time.Hour` and
synchronizing on a `secondRunStarted` channel instead of polling
`requestCount`, so the test fails without the wake instead of relying on
the 10ms ticker.
Closes https://github.com/coder/internal/issues/1500
> Mux is acting on Mike's behalf.
When switching between existing `/agents` chats, unsent file attachments
were kept only in React state and were lost when the chat page
remounted.
This adds chat-scoped draft attachment persistence for the compose
input, keeps edit-mode attachments isolated from compose drafts, warns
when a file cannot be saved to localStorage, and cleans up restored
draft payloads after upload or send.
> Mux is acting on Mike's behalf.
> 🤖 This PR was modified by Coder Agents on behalf of Jake Howell.
Fixes the `outsideBox` variant styling for tabs and simplifies the kebab
overflow logic. The overflow calculation now accounts for `column-gap`
between tabs so the menu trigger appears at the correct breakpoint.
- Fix Tailwind hover selector syntax for `outsideBox` variant
(`[&_[data-slot=tabs-trigger]:hover]` instead of
`[&_[data-slot=tabs-trigger]]:hover`)
- Account for `column-gap` in `useKebabMenu` overflow calculation via a
new `getTabGap` helper
- Use content-box width (`getContentBoxWidth`) for the initial overflow
pass so it matches `ResizeObserver`'s `contentRect.width`
- Consolidate `calculateTabValues` into a single-pass loop, removing the
separate `findFirstTabIndex` function
- Drop `FC` wrapper in favor of inline prop destructuring across all tab
components
- Forward `ref` through `TabsList` so consumers can attach refs directly
- Add `useKebabMenu` unit tests covering both all-visible and overflow
scenarios
Anthropic can reject replayed chat histories when a provider-executed
tool call, such as `web_search`, is present without its matching
provider result block.
This sanitizes unpaired Anthropic provider-executed tool calls during
prompt reconstruction, before Anthropic requests, and before persistence
so existing poisoned histories can continue and new malformed turns are
not stored.
Resolves: CODAGT-259
> Mux is acting on Mike's behalf.
## Problem
The CLI does not honor `default` values on template parameters in two
ways:
1. **`--use-parameter-defaults` rejects empty-string defaults.** The
check `parameterValue != ""` means `default = ""` in Terraform falls
through to an interactive prompt. In CI this causes an EOF error.
2. **`--use-parameter-defaults` only exists on `coder create`.** The
`start`, `update`, and `restart` commands never wire it through. SSH
auto-start passes empty `workspaceParameterFlags{}`, so users SSH-ing
into a stopped workspace with new template parameters get stuck in an
interactive prompt they cannot complete.
## Fix
### 1. Fix empty-string default detection and expose flag on all
commands
Replace `parameterValue != ""` with a check based on `!tvp.Required`. A
parameter with `Required==false` always has a valid default in
Terraform, even if that default is `""`. Also respect CLI defaults
provided via `--parameter-default`.
Move `--use-parameter-defaults` from a standalone option on `create`
into the shared `workspaceParameterFlags` struct. This exposes the flag
(and `CODER_WORKSPACE_USE_PARAMETER_DEFAULTS`) on `start`, `update`, and
`restart` via `allOptions()`. Wire it through
`buildWorkspaceStartRequest` so the resolver receives it.
### 2. SSH auto-start always uses defaults
Set `useParameterDefaults: true` on both `startWorkspace` calls in the
SSH auto-start path (initial start and the forbidden/upgrade fallback).
SSH is non-interactive and should never prompt.
Fixes https://linear.app/codercom/issue/DEVEX-180
Fixes https://github.com/coder/coder/issues/22272
<details><summary>Implementation notes</summary>
### Scoping decisions
- **`--yes` does not imply `--use-parameter-defaults`**: Making `--yes`
auto-accept defaults exposes a validation gap in the dynamic parameter
path (client-side validation happens during prompting, and skipping
prompts bypasses it). This is deferred to a follow-up that also
addresses `codersdk.ValidateWorkspaceBuildParameter` integration in the
resolver. Tracked in PLAT-114.
- **Explicit overrides always win**: `--parameter`,
`--rich-parameter-file`, and `--preset` are resolved in stages 1-5 of
the resolver, before `resolveWithInput` runs. No change needed for
precedence.
- **`!tvp.Required` vs `parameterValue != ""`**: The `Required` field is
set by the Terraform provider based on whether a `default` is present.
This is the canonical signal for "has a default," not the string value
itself.
</details>
> Generated with [Coder Agents](https://coder.com/agents)
Bumps the `charm.land/fantasy` replace directive to pick up
<https://github.com/coder/fantasy/pull/29>, which adds `gpt-5.5` and
`gpt-5.5-pro` to `responsesReasoningModelIDs`.
Without this, chatd's `useOpenAIResponsesOptions` returns false for
GPT-5.5, so it falls back to Chat Completions and never attaches
`ResponsesProviderOptions` (losing `store=true` + `previous_response_id`
chaining and other Responses-only features).
## Changes
- `go.mod`: `github.com/coder/fantasy
v0.0.0-20260416152503-959aa39579d2` →
`v0.0.0-20260424191546-5ab464a305f4`
- `go.sum`: updated hashes
Verified `go build ./coderd/x/chatd/...` passes locally.
Created on behalf of @ibetitsmike
Co-authored-by: blink-so[bot] <211532188+blink-so[bot]@users.noreply.github.com>
Once a user has touched a field, it is better to leave it alone and display explicit validation errors over silently overwriting their inputs. Same for auto-filled values (whether from query parameters or a previous build).
Previously, `chats.last_model_config_id` was not updated when a user
sent a mid-chat message with a different model, and queued messages did
not store their own per-turn model, so promotion ran against whatever
the chat row said at promote time. Chat watch events also did not merge
`last_model_config_id` into the site's root, child, and per-chat
caches, so sidebar labels stayed stale after direct sends and queued
promotions.
- Add nullable `chat_queued_messages.model_config_id`, backfilled from
`chats.last_model_config_id`. Queued inserts round-trip the effective
model id at enqueue time.
- In `coderd/x/chatd`, direct sends update `chats.last_model_config_id`
inside the same transaction that inserts the admitted user message.
Manual promotion and auto-promotion use the queued row's stored
`model_config_id`, with a fallback to `chats.last_model_config_id`
for legacy NULL rows during rollout.
`PromoteQueuedOptions.ModelConfigID`
is now ignored.
- On the site, extract `mergeWatchedChatSummary` and
`mergeWatchedChatIntoCaches` in `site/src/api/queries/chats.ts` so
status-change watch events merge `last_model_config_id` into the
root infinite chat list, the parent-embedded child entry, and the
per-chat `chatKey(chatId)` cache. `updated_at` guards against stale
watch payloads clobbering newer cached state, while diff status
events still merge their PR metadata because they are timestamped
outside the chat row. Watch timestamps are compared as instants so
variable fractional precision does not make fresh events look stale.
- Queued promotion validates stored model config IDs before admission.
Invalid legacy queued IDs fall back to the chat's current model config
instead of dropping the queued message during auto-promotion.
- Backend and frontend regression coverage added for admission, queue
promotion (including FIFO across mixed models, legacy NULL fallback,
and invalid queued model IDs), and chat watch cache merging.
> Mux is acting on Mike's behalf.
*Disclaimer: implemented by a Coder Agent using Claude Opus 4.6*
Follow-up to https://github.com/coder/coder/pull/24630 which added
backend Charm Crush client detection.
## Summary
Adds a `"Charm Crush"` case to the `AIBridgeClientIcon` switch so the
client filter and request logs show an icon instead of the generic
question mark fallback.
Uses a 💘 emoji-based SVG as a placeholder, matching Crush's branding.
Replace with an official brand SVG when one becomes available from the
Charm team.
## Changes
- `site/static/icon/charm-crush.svg`: new emoji-based placeholder icon.
-
`site/src/pages/AIBridgePage/RequestLogsPage/icons/AIBridgeClientIcon.tsx`:
added `"Charm Crush"` case.
Rolldown's tokio workers stall when competing with Go compilation
and the production site build for CPU, causing Vite transform
requests to hang. Vitest browser mode has no import-phase timeout,
so a stalled browser import() blocks the run indefinitely.
Adds a deployment-wide admin override for general delegated subagents.
## What changed
- store the general override in `site_configs` and expose it through the
shared `agent-model-override/{context}` API
- apply the general override when spawning delegated general subagents,
while preserving the existing Explore override behavior
- reuse a shared Agents settings form for the general and Explore
override sections
## Validation
- `make gen`
- `go test ./coderd -run 'TestChatModelOverrides'`
- `go test ./coderd/x/chatd -run
'TestSpawnAgent_(GeneralUsesConfiguredModelOverride|GeneralOverrideLogsAndFallsBackWhenCredentialsUnavailable|GeneralOverrideLogsAndFallsBackWhenProviderDisabled)'`
- `pnpm -C site lint:types`
- `pnpm -C site test:storybook --
AgentSettingsAgentsPageView.stories.tsx`
- `make lint`
- `make pre-commit`
> Mux is acting on Mike's behalf.
- **computeruse.go**: Decode base64 screenshot data before storing in
`ToolResponse.Data` (was casting base64 string to bytes without
decoding)
- **chatloop.go**: Re-encode `ToolResponse.Data` to base64 via
`base64.StdEncoding.EncodeToString` instead of `string()` cast
- **mcpclient.go**: UTF-8 validate all text from MCP responses in
`convertCallResult()` using `strings.ToValidUTF8`
- **chatprompt.go (persist)**: Defense-in-depth UTF-8 sanitization of
text and media Text fields before database storage
- **chatprompt.go (replay)**: Antivenom layer that validates base64 and
UTF-8 at read time, auto-healing already-poisoned chats without
requiring a migration
- `TestToolResultAntivenom`: 4 subtests covering poisoned text, poisoned
media, valid media round-trip, and media with invalid UTF-8 text
- Adds `TestConvertCallResult_UTF8Sanitization`: 4 subtests covering invalid
UTF-8 in TextContent, EmbeddedResource, valid passthrough, and
multi-part
- Adds `TestComputerUseTool_Run_ScreenshotDataIsDecodedBinary`: Verifies no
double-encode in the computer-use path
- Updated existing computer-use tests for the new decoded-binary
contract
> 🤖
The UI already prevents child (delegated/subagent) chats from being
pinned, but the `PATCH /api/experimental/chats/{chat}` endpoint did not
enforce this. A direct API call could pin a child chat.
- Add a `400 Bad Request` guard in `patchChat` when `pinOrder > 0` and
the chat has a `ParentChatID`
- Add `TestChatPinOrder/RejectsChildChat` test
> 🤖
Explore sub-agents previously could not use `web_search` or external MCP
tools. `runChat` hard-skipped both for Explore. Lifting those guards
naively would over-grant tools, because a child chat could outlive the
spawning turn's plan-mode filter.
This change persists the spawning parent turn's filtered external MCP
server IDs onto the child Explore chat, and simplifies the Explore
provider-tool filter in `runChat`:
- New `resolveExploreToolSnapshot` helper: computes the child's
inherited external MCP subset by running the parent's configs through
`filterExternalMCPConfigsForTurn` (plan-mode policy) and, if the parent
is itself an Explore child, further narrowing to the parent's own
persisted `MCPServerIDs`. The result is written to the child's
`MCPServerIDs` column at spawn time.
- The existing `mcp_server_ids` column is the sole durable snapshot. No
new chat column is added.
- `runChat` for Explore children: loads MCP tools from the persisted
snapshot, and keeps only `web_search` from provider-native tools (to
block computer-use and other write-style tools, since Explore is
read-only). Whether `web_search` is actually available is a per-model
decision, determined by the current model config, just like a main chat.
- Built-in Explore allowlist is unchanged. Workspace-local MCP remains
excluded for Explore.
Verification: `go build ./...`, `go test ./coderd/x/chatd/... -count=1`,
`make gen` (clean tree), `make lint/emdash`, `go vet`. Deep-review ran
12 reviewers on the feature and 5 on the clarity refactor; CAR reviewed
and approved; a subsequent scope reduction dropped a temporary
`allow_web_search` column in favor of per-model handling.
> Mux is acting on Mike's behalf.
The agents-access role previously granted chat permissions at user
scope, but chats are org-scoped objects. Rego skips user-level perms
when org_owner is set, making the grants invisible. Handler-level
band-aids used synthetic non-org-scoped objects as a workaround.
- Migrates agents-access from users.rbac_roles (site-level) to
organization_members.roles (org-scoped) via DB migration
- Redefines agents-access as a predefined org-scoped builtin role
alongside organization-admin, organization-auditor, etc., with
Member permissions granting chat create/read/update
- Excludes ResourceChat from OrgMemberPermissions so org membership
alone no longer grants chat access
- Fixes handler Authorize checks to use org-scoped objects with
semantically correct actions (ActionUpdate for message/tool operations)
- Grants org admins the ability to assign agents-access
Closes#24250
Fixes CODAGT-174
Note: this does not update the "Usage" endpoints. Tracked by CODAGT-161.
> 🤖
`OpensAdminSubPanelOnMobile` story has been failing due to not finding a
role="link" with name "Settings".
Agent investigated and claimed the Settings icon gets hidden in mobile
view.
Suggested fix is to start the story directly at `/agents/settings`.
> Generated with [Coder Agents](https://coder.com/agents)
Archived chats accept mutations (messages, edits, queued-message
promotions, tool-result submissions) via the API, causing them to
re-enter the processing pipeline. This violates the hard-stop
design intent from PR #23758.
Add archived checks at three layers:
- HTTP handlers (postChatMessages, patchChatMessage,
promoteChatQueuedMessage, postChatToolResults): return 400
after auth so callers get a clear error.
- Daemon functions (SendMessage, EditMessage, PromoteQueued,
SubmitToolResults): return ErrChatArchived after row lock,
guarding against future callers that bypass the handler.
- AcquireChats SQL: filter out archived chats so they are never
acquired for processing.
Fixes CODAGT-245
Michael Suchacz