A chat that fails generation with a context overflow (for example `Input
length 262625 exceeds the maximum allowed input length of 262112
tokens`) is stuck in a catch-22: `POST /chats/{id}/compact` returns 409
because the `RequestCompaction` transition is only allowed from the
waiting state, and the only other way out of the error state is sending
or editing a message, which re-runs generation with the same oversized
prompt and fails again. Compaction is exactly the recovery a
context-overflowed chat needs, and it is unreachable exactly when it is
needed.
Three semantic changes:
- Allow `RequestCompaction` from the error states: `E0 -> R0` and `E1 ->
R1` (queued messages are preserved and processed after the compaction
turn).
- Clear `last_error` in `Tx.RequestCompaction`, matching the
architecture rule that transitions leaving `E0`/`E1` clear the stored
error. Without this a successful compaction would land in waiting with a
stale persisted error.
- Grant the compaction turn a fresh history epoch: a
`grant_history_epoch` flag on `UpdateChatExecutionState` sets
`history_version = snapshot_version`, resets `generation_attempt`, and
clears `retry_state` in the same atomic update that clears `last_error`
(mirroring the `chat_messages` trigger postcondition). The transition
inserts no history, so without this the turn inherits the failed turn's
spent retry budget, and resetting the counter alone could collide with
message part episode keys still retained on the erroring replica.
No frontend change is required: the chat input is already enabled in the
error state and `/compact` submission already handles both the success
and 409 paths. Also updates ARCHITECTURE.md (transition matrix,
endpoint, and manual compaction sections), the endpoint's swagger
description, and SDK comments.
> Mux created this PR on Mike's behalf.
<!-- mux-attribution: model=claude-sonnet-4-6 thinking=high -->
Adds a user-triggered `/compact` action for Coder Agents chats: typing
`/compact` in the composer (or picking it from the `/` trigger menu)
summarizes the conversation so far to free up context window space.
## How it works
- New `POST /api/experimental/chats/{chat}/compact` endpoint
(owner-only, RBAC `ActionUpdate`, excluded from the public API reference
via `x-apidocgen skip`). It marks the chat with a durable one-shot
`chats.compaction_requested_at` signal and moves it `waiting -> running`
via a new `RequestCompaction` state transition; no message row is
inserted. AI Gateway attribution needs no per-request key: generation
preparation resolves the owner's synthetic API key (#27170) like any
other turn.
- `RequestCompaction` hands off chat ownership (clears
`worker_id`/`runner_id`) so a worker acquisition hint is published;
since the transition changes no history, the previous runner could
otherwise miss the request under reordered pubsub delivery.
- The background chat worker picks the chat up like any other turn. A
pending manual request takes precedence over turn completion in the
generation decision, and forces compaction even below the automatic
threshold (and when compaction is disabled via threshold=100). The
commit step consumes the request marker in the same transaction; any
transition that ends the turn clears stale markers.
- The summary triplet reuses the automatic-compaction path, now tagged
with a `source` (`automatic` | `manual`) that is plumbed through
streamed progress parts, persisted tool JSON, and the UI label
("Summarized (manual)").
- Validation order: busy chats reject with 409 (state-machine conflict),
empty/already-compacted chats with 409 "nothing to compact", archived
chats with 400; the owner usage-limit check runs last so no-op requests
surface the specific conflict instead of a limit error.
- Web UI: the `/` trigger menu now has a built-in "Commands" group
listing `/compact`; submit intercepts exactly `/compact` and calls the
endpoint instead of sending a message. A personal or workspace skill
named `compact` takes precedence over the built-in command; while skill
collisions are still resolving, an exact `/compact` submission is
blocked with a retryable hint instead of leaking as message text.
History and queued-message edits are never intercepted. After
compaction, the context usage indicator resets to its unknown state
until the next assistant response reports fresh usage, instead of
showing the stale pre-compaction number.
- codersdk: `ExperimentalClient.CompactChat`.
Worker-path execution (rather than compacting synchronously in the
handler) reuses the existing lock fencing, live "Summarizing..."
streaming, retry accounting, restart resilience, and debug-run
observability. Rationale documented in `coderd/x/chatd/ARCHITECTURE.md`.
## Testing
- State machine: transition-matrix coverage for `RequestCompaction`,
marker lifecycle tests (carried by lease renewals/queue appends, cleared
by terminal transitions, consumed by commit), ownership handoff +
acquisition hint assertions.
- Worker: decision-ordering and forced-compaction unit tests;
active-server end-to-end test (manual compact below threshold produces a
`source=manual` summary, returns to `waiting`, no assistant follow-up;
busy chat rejected).
- API: success, archived, non-owner, RBAC-denied, empty-chat, no-daemon
cases; usage-limit ordering (at-limit owners still get
state/nothing-to-compact conflicts for no-op requests, with marker
rollback).
- Frontend: Storybook play tests for the Commands menu group, submit
intercept, skill-name collision, queued-edit passthrough, and
manual/automatic tool rendering; unit tests for command availability
resolution and the post-compaction context usage reset.
> This PR was created by Mux, an AI coding agent, working on Mike's
behalf.
When a workspace has no POSIX sh on PATH (typical for fresh Windows workspaces), the execute tool fails with a raw `exec: "sh": executable file not found in %PATH%` error the model cannot act on.
This change:
- Enriches the above error in chattool with remediation steps and a docs link.
- Documents the requirement in the Coder Agents architecture page.
> This PR was generated by Coder Agents on behalf of @johnstcn
Tool errors caused orchestrators to abandon spawned agents. Bare error
responses and the close_agent name framed delegation as one-shot: one
transient failure or timeout ended the work, and the orchestrator had no
way to recover or reuse agents.
Renames close_agent to interrupt_agent with a hidden backward-compatible
alias. wait_agent and message_agent return structured payloads instead
of bare errors, so the orchestrator can retry after a timeout, recover
from an error status, or redirect an idle agent. Adds list_agents so
orchestrators can rediscover spawned agents. Adds root-only
orchestration guidance for error recovery.
Closes coverage gaps in `docs/ai-coder/agents/` and aligns nav
references with the current UI (post #24574 Behavior split, post #24644
Insights removal).
**Content fixes:**
- Replace site-wide `coder users edit-roles` flow with org-scoped
`agents-access` role (per migration `000475`). CLI examples now preserve
existing org roles since `edit-roles` overwrites the full set.
- Correct computer-use claim: supports Anthropic *and* OpenAI providers,
configured under the Virtual desktop experiment.
- New `platform-controls/experiments.md` covering Virtual desktop,
Advisor, and Chat debug logging (each as: what, how to enable, API).
Includes the Debug tab in the chat right panel.
- Trim `models.md` "Model overrides" to essentials: two layers (admin
subagent, user personal), contexts table, resolution order, API pointer.
- Remove retired `platform-controls/pr-insights.md` (page + manifest +
cross-links).
**Nav cleanup:**
- Admin-only tabs use the full `Agents > Settings > Manage Agents >
<Tab>` path; user-side tabs keep `Agents > Settings > <Tab>`.
- Replace stale "Behavior" references with Instructions / Lifecycle /
Experiments to match the current sidebar.
- Replace references to the removed top-bar Admin dialog with the
Settings sidebar.
<details>
<summary>Decision log</summary>
- Experimental features were originally drafted as a standalone Advisor
page plus inline sections in `platform-controls/index.md`. Consolidated
into one `experiments.md` since no individual feature warrants a full
page yet and parallel short sections are easier to scan.
- Reviewer feedback on early drafts: drop the inline experiments list
from `index.md` (avoid drift), drop the "users created before this role
was introduced" note (handled transparently by migration `000475`),
specify the full nav path for per-model pricing, link the
`type=computer_use` row in `architecture.md` to the Experiments page.
- CLI bulk-grant script previously called `edit-roles <user>
agents-access`. That replaces the user's full org role set, so the
script would silently strip `organization-admin`,
`organization-template-admin`, etc. Rewrote to read each user's current
roles, append `agents-access`, dedupe, and write the union back.
</details>
PR generated with Coder Agents.
PR #24772 (merged 2026-05-04) added OpenAI alongside Anthropic for
computer use, plus an admin selector under the virtual desktop toggle.
Three places in the agents docs still said "Anthropic only" — this
updates them.
No other content changes. Anthropic is still the default.
Fixes
[CODAGT-310](https://linear.app/codercom/issue/CODAGT-310/enable-openai-computer-use-in-codercoder)
---
@nickvigilante — heads up, the kind of release-train drift we keep
hitting:
- Feature is on `main`, so docs on `main` need to describe it.
- Feature is **not** in `release/2.33` and **not** in `v2.34.0-rc.0`
(both cut before #24772 merged). It will ship in v2.34.
- `coder.com/docs` follows `main`, so once this lands, v2.33 users see
"OpenAI is supported" and find no toggle.
Fwiw our [`doc-check`
workflow](https://github.com/coder/coder/blob/main/.github/workflows/doc-check.yaml)
would have caught this on #24772 — it's exactly what it's for. It [did
trigger](https://github.com/coder/coder/actions/runs/25326759671) but
the chat-create step errored out (curl exit 22) and nobody re-ran it, so
the analysis never happened. Worth tightening that path so a transient
API blip doesn't silently skip the check.
> Generated with [Coder Agents](https://coder.com/agents)
Unify the three subagent spawn tools (`spawn_agent`,
`spawn_explore_agent`, `spawn_computer_use_agent`) behind a single
`spawn_subagent` tool keyed by a `subagent_type` discriminant
(`general`, `explore`, `computer_use`). Mirrors the single-entry-point
pattern already used by `task` in mux while keeping `wait_agent`,
`message_agent`, and `close_agent` as separate lifecycle tools.
A new backend subagent definition catalog
(`coderd/x/chatd/subagent_catalog.go`) is the source of truth for tool
description, prompt guidance, availability rules (plan mode,
desktop/Anthropic gating), and child-chat option building.
`spawn_subagent` advertises only the types available in the current
context and validates `subagent_type` server-side; context inheritance
still flows through the existing `createChildSubagentChatWithOptions`
path. `wait_agent`, `message_agent`, and `close_agent` responses now
include a server-derived `subagent_type` so the UI stops inferring
lifecycle state from tool names.
The frontend gets a shared normalization helper
(`site/src/pages/AgentsPage/components/ChatElements/tools/subagentDescriptor.ts`)
that maps either legacy tool names or new `spawn_subagent` args into a
common descriptor (action, variant, icon, fallback copy). Legacy
transcripts still render identically; `Tool.tsx`, `SubagentTool.tsx`,
`ToolLabel.tsx`, `ToolIcon.tsx`, and `messageParsing.ts` now key off the
descriptor instead of hard-coded names. Existing UI copy is preserved
(`Spawning Explore agent...`, `Using the computer...`, computer-use
monitor icon and Open Desktop affordance).
> This PR was opened by Mux working on Mike's behalf.
## Summary
Allow root plan-mode chats to use MCP tools from external servers that
an admin has explicitly approved for plan mode. Workspace MCP and
plan-mode subagents remain blocked.
## Problem
`chatd.go` excluded every MCP tool when `isPlanModeTurn` was true, so
planning had no access to tools like docs search, ticketing, etc.
Lifting that guard wholesale was unsafe: `mcp_server_configs` already
has centralized admin governance, but workspace-local MCP (discovered
from agent `.mcp.json`) does not, and subagents use a narrower trust
boundary.
## Fix
Add an admin-controlled per-server `allow_in_plan_mode` flag (default
`false`) and gate plan-mode MCP access on it.
### Backend / schema
- New migration `000472_mcp_server_allow_in_plan_mode.{up,down}.sql` and
matching fixture update.
- `mcpserverconfigs.sql` + generated code: persist and read the new
column.
- `codersdk/mcp.go`: thread the field through `MCPServerConfig`,
`Create*`, and `Update*` request types.
- `coderd/mcp.go`: validate, persist, and return the flag in
get/list/create/update handlers.
### chatd
- `coderd/x/chatd/chatd.go`: pre-filter selected external MCP configs by
`AllowInPlanMode` before calling `mcpclient.ConnectAll` on plan-mode
root turns. Workspace MCP discovery is skipped entirely on plan-mode
turns.
- Single helper decides whether a tool is available in plan mode, used
both at construction and for active-tool filtering (defense in depth).
Plan-mode subagents, dynamic tools, provider-native tools, computer-use,
and workspace MCP stay unchanged.
- `coderd/x/chatd/prompt.go`: update the root plan-mode overlay text to
match the new boundary.
### UI
- `MCPServerAdminPanel.tsx`: add an explicit toggle ("Allow all tools
from this MCP server in root plan mode") next to the existing governance
controls.
- Regenerated `site/src/api/typesGenerated.ts`.
### Docs
- `docs/ai-coder/agents/architecture.md`: replace the blanket "MCP is
unavailable in plan mode" note with the new root-only, external-only,
admin-approved policy. Explicitly call out that workspace MCP and
plan-mode subagents are still excluded.
### Tests
- Plan-mode visibility (approved vs non-approved external server).
- Plan-mode invocation of an approved external MCP tool.
- End-to-end plan-mode workflow that uses an approved MCP tool and then
reaches `propose_plan`.
- Regressions: workspace MCP still excluded in plan mode; plan-mode
subagents still on the restricted tool boundary; existing tool
allow/deny list filtering still applies.
## Policy precedence
`allow_in_plan_mode` is an **additional** requirement on top of existing
`enabled`, availability, chat-selected / forced server IDs, and tool
allow/deny lists. It approves **all tools on that server** for root plan
mode; a per-tool plan allowlist is deliberately deferred.
## Follow-ups (explicitly out of scope)
- Whether plan-mode subagents should inherit approved external MCP
tools.
- Workspace-local MCP safety model (agent-side `.mcp.json` schema vs. a
coderd-managed workspace MCP config).
## Validation
- `go vet ./coderd/x/chatd/...`
- `go test ./coderd/x/chatd -run 'TestPlan.*|TestMCP.*' -count=1`
- `go test ./coderd/x/chatd -count=1 -timeout 5m` (full chatd suite)
- `make fmt` (no diff)
> Mux opened this PR on Mike's behalf.
> This PR was authored by Mux on behalf of Mike.
## Summary
- add persistent plan mode for chats and the chat-specific plan file
flow
- add structured planning tools such as `ask_user_question` and
`propose_plan`
- keep `write_file` and `edit_files` constrained to the chat-specific
plan file during plan turns
- allow shell exploration in plan mode, including subagents, via
`execute` and `process_output`
- block implementation-oriented, provider-native, MCP, dynamic, and
computer-use tools during plan turns
- update the chat UI, tests, and docs for the new planning flow
Renames the "Security implications" section to "Security posture" and
reframes the intro paragraph. "Implications" reads as a caveat or
warning; the section actually describes built-in structural guarantees
of the control plane architecture.
> PR generated with Coder Agents
Fixes several documentation gaps and inaccuracies in the Coder Agents
docs identified during a deep review against the current product state.
## BYOK (User API Keys)
`models.md` stated *"Developers cannot add their own providers, models,
or API keys"* — this has been incorrect since the provider key policy
system shipped (Apr 2, #23751/#23781).
- Added **Key policy** section documenting the three admin toggles
(`central_api_key_enabled`, `allow_user_api_key`,
`allow_central_api_key_fallback`) with a truth table showing all
resolution outcomes
- Added **User API keys (BYOK)** section covering the developer-facing
key management page, status indicators, selection priority, and key
removal
- Updated `platform-controls/index.md` to reference BYOK instead of
claiming keys are admin-only
## Reasoning effort enum fixes
- **OpenAI**: removed `none` — code accepts `minimal, low, medium, high,
xhigh`
- **OpenRouter**: narrowed to `low, medium, high` per
`ReasoningEffortFromChat` in `chatprovider.go`
## Tool table completeness
- Added `spawn_computer_use_agent`, `read_skill`, `read_skill_file` to
`index.md` tool table
- Added "Workspace extension tools" section to `architecture.md` for
`read_skill`/`read_skill_file`
- Fixed orchestration restriction note to list all 5 gated tools instead
of just `spawn_agent`
- Added conditional availability notes for desktop and skills tools
## Platform controls
Three admin-only settings existed in the Behavior tab with no
documentation:
- **Virtual desktop** — admin toggle, Anthropic + portabledesktop
requirements
- **Workspace autostop fallback** — default TTL for agent workspaces
without template-defined autostop
- **Data retention** — moved `chat-retention.md` into
`platform-controls/` since it's admin-only, fixed nav path
---
> PR generated with Coder Agents
Fixes factual errors found during a review of all pages under
`/docs/ai-coder/agents/`.
## Tool tables (`index.md`, `architecture.md`)
Both pages had incomplete tool tables. Added:
- `process_output`, `process_list`, `process_signal` — core workspace
tools always registered alongside `execute`, missing from both pages
- `propose_plan` — platform tool (root chats only), missing from both
pages
- `spawn_computer_use_agent` — orchestration tool (conditional), missing
from architecture.md
Also fixed the architecture.md claim that the agent is "restricted to
the tool set defined in this section" — it now mentions skills and MCP
tools with links to the relevant pages.
## Model options (`models.md`)
- **OpenAI / OpenRouter Reasoning Effort**: docs listed `low`, `medium`,
`high` — code has `none`, `minimal`, `low`, `medium`, `high`, `xhigh`.
Fixed both.
- **Removed hidden fields** that never appear in the admin UI:
- Google: Safety Settings (`hidden:"true"`)
- OpenRouter: Provider Order, Allow Fallbacks (parent struct
`hidden:"true"`)
- Vercel: Provider Options (`hidden:"true"`)
---
*PR generated with Coder Agents*
The tool descriptions pushed agents toward backgrounding anything over
5 seconds, including builds, tests, and installs where you actually
want to wait for the result. This led to unnecessary process_output
round-trips and missed the foreground timeout-to-reattach workflow
entirely.
Reframe background mode as the exception (persistent processes with
no natural exit) and foreground with an appropriate timeout as the
default. Replace "background process" with "tracked process" in
process_output, process_list, and process_signal since they work on
all tracked processes regardless of how they were started.
Addresses five documentation gaps identified from an internal agents
briefing Q&A, specifically around what permissions an agent inherits
from the user:
1. **No privilege escalation** — Added explicit statement that the agent
has the exact same permissions as the user. No escalation, no shared
service account.
2. **Cross-user workspace isolation** — Added statement that agents
cannot access workspaces belonging to other users.
3. **Default-state warning** — Added WARNING callouts that agent
workspaces inherit the user's full network access unless templates
explicitly restrict it.
4. **Tool boundary statement** — Added explicit statement that the agent
cannot act outside its defined tool set and has no direct access to the
Coder API.
5. **Template visibility scoped to user RBAC** — Clarified that template
selection respects the user's role and permissions.
Changes across 3 files:
- `docs/ai-coder/agents/index.md`
- `docs/ai-coder/agents/architecture.md`
- `docs/ai-coder/agents/platform-controls/template-optimization.md`
---
PR generated with Coder Agents
Michael Suchacz