Commit Graph
775 Commits
Author SHA1 Message Date
Mathias Fredriksson 1aff810443 chore: sync fantasy with upstream v0.45.5 (#31288)
Moves the fantasy pin to the tip of `coder_2_33` (`85ffbbe`). That
commit is coder/fantasy#72, which merges 87 upstream
charmbracelet/fantasy commits, v0.40.0 to v0.45.5. Between the two pins
117 files change. Upstream's own v0.40.0 to v0.45.5 change, without
tests, testdata, examples, `go.sum` and the kronk provider, is 43 files,
+2271/−928.

## What reaches chatd

chatd builds only the `anthropic`, `openai` and `openaicompat` fantasy
clients (`chatprovider.ModelFromConfig`); other vendors reach it through
the AI Gateway on one of those. So the upstream changes to the Google,
Vercel, OpenRouter, Bedrock, Azure and kronk providers, and to the
`fantasy.Agent` loop and its retries, do not run in chatd. These do:

- **Chat Completions and openaicompat streams** keep reasoning on
batched and DeepSeek-shaped chunks, keep `finish_reason: length` when
tool calls are truncated, and stop sending empty reasoning fields.
openaicompat drops assistant turns that hold only reasoning.
- **Token usage.** When an OpenAI-shaped backend reports reasoning
tokens apart from completion tokens (reasoning larger than output),
output tokens now include them, so that backend's reported output tokens
and cost rise. OpenAI's own shape, where reasoning is a subset, is
unchanged.
- **Responses.** A content-filtered turn no longer reads as a tool-call
turn.
- **Context overflow.** Baseten, Fireworks and Parasail prompt-length
errors set `ContextTooLargeErr`, which compaction acts on.
- **Anthropic.** Tool calls whose content block never closes are
reported. For Claude 5 models, thinking defaults to `display:
"summarized"` when the chat config sets none; a configured display still
wins. `TestThinkingDropBlockModel_RetriesWithEffort` now expects the
field in the retry body.
- **Anthropic keepalives.** Fantasy forwards `message_start`,
`message_delta` and unknown events as content-free `keepalive` parts.
The first one arrives before any model output, so this PR makes
`ttft_seconds` and the `time_to_first_token` stage skip them, as they
skip warnings. They still reset the stream silence guard.
`processStepStream` ignores them, and the debug recorder counts them in
its part count but records no content.
- **Tool-result media** now goes through upstream's shared
`ToolRunBuffer`, which replaces coder/fantasy#64 and does the same
thing.

#72 keeps coder's anthropic-sdk-go and openai-go forks, Go 1.26 and
kronk v1.31.5, so upstream's SDK and Go moves do not come along.

`go mod tidy` raises the modules fantasy v0.45.5 requires, among them
otel exporters v1.47, the AWS SDK, `google.golang.org/api` v0.307 and
go-openapi v1. All are upgrades.

The `go.mod` comment now names #72 as the upstream sync.

coder/coder#29861 stacks on this PR to pin coder/fantasy#62.

<details>
<summary>Verification</summary>

- `go build ./...`.
- `go test` passes on every package that imports fantasy
(`coderd/x/chatd/...`), and `go test ./coderd -run
'TestChat|TestExpChat|TestMCP|TestAIBridge'` passes.
-
`TestGuardedStreamTTFTStage/WarningsAndKeepalivesThenContentObservesOnce`
fails without the TTFT change (TTFT observed 0s instead of 1s) and
passes with it.
- `make pre-commit` passes. The generated Prometheus docs pick up the
metric help change.

Not verified against live backends in this PR: the stream fixes, the
token usage change and the context-overflow patterns have no coder test
against a real provider. coder/fantasy#72 reports a local chatd run
against the AI Gateway on Anthropic, Responses, OpenAI Chat and
openaicompat.

</details>

> 🤖 This PR was created with the help of Coder Agents, and _will be_
reviewed by a human. 🏂🏻
2026-10-27 11:29:10 +11:00
Jon Ayers 688c4a390c fix: bump Go to 1.26.10, terraform to 1.16.7, alpine to 3.24.3 (CVE-2026-111943) (#31301)
Mitigates CVE-2026-111943 (compromised eMudhra intermediate CAs).

- **Go 1.26.9 to 1.26.10** (`go.mod`, `mise.toml`, `mise.lock`). Go
1.26.10 no longer trusts certificates issued through the compromised
intermediates.
- **Bundled Terraform 1.16.6 to 1.16.7** (`install.go`, `install.sh`,
`scripts/Dockerfile.base`, `mise.toml`, `mise.lock`, `flake.nix`, airgap
docs, `testdata/generation.sha1`).
[1.16.7](https://github.com/hashicorp/terraform/releases/tag/v1.16.7) is
built with Go 1.26.10. Checksums come from the GPG-verified
`terraform_1.16.7_SHA256SUMS`. `maxTerraformVersion` stays at 1.16.9.
- **Base image `alpine:3.24.2` to `alpine:3.24.3`**. 3.24.2 ships
`ca-certificates-bundle-20260909-r0`; 3.24.3 ships `20261020-r0`.

Verification:
- Rebuilt `scripts/Dockerfile.base` for linux/amd64 and linux/arm64.
Both contain `ca-certificates-bundle-20261020-r0` and `Terraform
v1.16.7`.
- `go test ./provisioner/terraform/...` passes with Terraform 1.16.7.
- `scripts/check_go_versions.sh` passes.
- Local git hooks were skipped at the author's request because the
dogfood host was overloaded. CI is the gate.

Backports to `release/2.34` through `release/2.38` follow as separate
PRs. Older Terraform minor lines did not get the Go 1.26.10 rebuild
(1.15.9 is built with Go 1.25.10, 1.14.9 with Go 1.25.8), so
release/2.35 also moves to 1.16.7.

<details>
<summary>Decision log</summary>

- The request named Go 1.26.1. Main and every release branch were
already on 1.26.9, so @sreya confirmed the target is 1.26.10.
- release/2.34 stays on the Alpine 3.23 line (`alpine:3.23.7`, which
also ships `ca-certificates-bundle-20261020-r0`) instead of jumping to
3.24.
- release/2.35 had drifted: `install.go` pinned 1.14.5 (max 1.14.9)
while the image and `mise.toml` used 1.15.5. Its backport sets both to
1.16.7 and raises max to 1.16.9.

</details>

> Generated by Coder Agents on behalf of @sreya.
2026-10-26 16:47:25 -07:00
Nick Vigilante 9edb2e388d feat: enable Coder.USSpelling Vale rule (#30790)
Depends on #30782.

Adds `Coder.USSpelling` at `warning` severity.
The rule flags an `-ise` family (`organise`, `initialise`,
`authorisation`) and a list of British spellings (`behaviour`,
`labelled`, `cancelled`, `analogue`, `whilst`, and others) and suggests
the US form.
It covers `docs/**` except `docs/.style/style-guide/**` and the
generated `docs/reference/{api,cli}/*.md` pages.

Baseline: 13 findings before cleanup, 0 after.
The first commit fixes 11 of them: 10 in hand-written pages and 1 in the
Prometheus metrics table, which comes from metric help text.
Two findings sit in generated reference pages, which `.vale.ini`
excludes for this rule: the CLI alias `favourite` and `amongst` in a Go
comment from a vendored dependency (the DERP map schema).
The metric help text comes from `querymetrics.go`, so the fix lands
there and in `generated_metrics` and `prometheus.md`.

Known false positive: Vale doesn't skip bold text, so a UI label in bold
that keeps British spelling is flagged.
The rule comment and the guide footer say to use a code span for such a
label.

Linear: DOCS-1280

<details>
<summary>Decision log</summary>

- **Rule type**: `substitution`, so each message names the US form.
The `-ise` family can't name a specific replacement, so it suggests "the
-ize form".
- **`cancelled`**: the guide's "Use US spelling" section lists
`canceled` and `canceling`.
The rule flags the double-l verb forms and leaves `cancellation` alone,
which Merriam-Webster spells with one `l`.
- **Exceptions**: Vale applies an exception to the matched word, so each
is anchored with `^` and `$` and case-insensitive.
The list covers `-wise` words and stems for `advise`, `advertise`,
`exercise`, `promise`, `premise`, `compromise`, `surprise`, `comprise`,
`concise`, `precise`, `revise`, `supervise`, `televise`, `devise`,
`improvise`, `franchise`, `merchandise`, `disguise`, `despise`,
`paradise`, `cruise`, `noise`, `raise`, `rise`, `wise`, and others, each
with the endings `e`, `es`, `ed`, `ing`, `er`, and `ers`.
  The `mise` tool is excepted by name.
- **Extra British spellings** beyond the guide table: other doubled-`l`
verbs (`signalled`, `levelled`), `metre`, `litre`, `fibre`, `theatre`,
`calibre`, `offence`, `pretence`, `analyse`, `judgement`,
`acknowledgement`, `ageing`, `artefact`, `sceptic`, `aluminium`,
`practise`, `enrol`, `fulfil`, `instil`, `skilful`.
  None appears in the corpus today.
- **Generated pages**: excluded for `docs/reference/{api,cli}/*.md`, as
`Coder.UnitSpacing` does.
Changing the `favourite` alias would change CLI behavior, and the
`amongst` comment isn't ours to edit.
- **Prometheus doc**: edited by hand with matching table padding because
`go run ./scripts/metricsdocgen` already rewrites the whole table on the
base branch, independent of this change.

Test cases, run with `vale --config` against fixtures that are not
committed, in LF and CRLF:

- Positive: `organise`, `initialise`, `authorise`, `customisation`,
`authorisation`, `organisations`, `Organised`, `initialising`,
`recogniser`, `customisable`, every explicit list entry, and capitalized
forms.
- Negative: `advise`, `exercise`, `otherwise`, `Enterprise`, `premise`,
`compromise`, `promise`, `surprise`, `precise`, `concise`, `expertise`,
`revising`, `supervising`, `improvising`, `raise`, `noise`, `cruise`,
`franchise`, `merchandise`, `disguise`, `despise`, `paradise`,
`appraise`, `sunrise`, `arise`, `liaise`, `clockwise`, `cancellation`,
`canceled`, `mise`.
- Skipped: words in code spans and fenced code blocks, such as
`cancelled`.
- LF and CRLF results are identical.

Checks: `vale --output=line docs/` reports 0 `Coder.USSpelling`
findings, `make lint/style-claims` passes, and the pre-commit hooks pass
on both commits.

</details>

*Generated by [Coder Agents](https://coder.com/docs/ai-coder/agents) on
behalf of @nickvigilante.*
2026-10-23 22:18:30 -07:00
Nick Vigilante 6e51c66a97 docs: give every table header cell text and fail lint on empty ones (#30879)
## Summary

axe reports `empty-table-header` on 116 generated CLI reference pages
and several hand-written pages: a header cell with no text is announced
by screen readers as an unnamed column. This PR gives every table header
cell in `docs/` text, and adds a lint check so empty header cells can't
come back.

## How to review this PR

- `docs/reference/cli/`: 116 files, these all generate identically, so I
would only spot check 2-3 of them and skip the rest
- `docs/*` outside of `docs/reference/cli/`: 6 files, normal review of
these
- `scripts`: 5 files, code review of the Bash and Go files to make sure
we're properly inserting headers into each table

## What changed

**Generator.** `scripts/clidocgen` wrote an empty header row (`| | |`)
over each option's metadata table (Type, Environment, YAML, Default). It
now writes `| Property | Value |`. Subcommand tables already had `Name |
Purpose` and are unchanged. Regenerated with `make
docs/reference/cli/index.md`: 116 files under `docs/reference/cli/`
change. The header rows change, and the body rows of each option table
are re-padded to the wider first column; the content is otherwise
identical. Every option on those 116 pages now shows a visible `Property
| Value` header row above Type, Environment, YAML, and Default, so this
is a visual change on the docs site as well as a screen reader one. The
CLI's own `--help` output doesn't use this template, so it's unchanged.

**Audit logs.** `docs/admin/security/audit-logs.md` already has `Field |
Tracked` headers on `main` (the generator writes them), so there was
nothing to change there; the live-site finding predates that.

**Hand-written tables.**
- `docs/admin/networking/port-forwarding.md`: the CORS matrix had three
empty spanning corner cells. The corner now reads `From user` and `From
workspace` (spanning the three row-header columns, beside the Alice/Bob
and Workspace rows) and `To user`, `To workspace`, `To app` (over the
three row-header columns). The grid is still 3 header rows by 7 columns.
The From cells have `scope="row"` and the other header-row cells
`scope="col"`, so the From labels don't become column headers for the
body's label columns; the body's user, workspace, and app labels are
`<th scope="row">`; and a `<caption>` says columns are the origin and
rows the destination.
- `groups-roles.md`, `airgap.md`, `scale-testing.md`,
`workspace-tags.md`: these four corner cells are already filled by
#30873. The new check would fail CI without them, so this PR carries the
same four lines, byte-identical to #30873 (`Permission`, `Component`,
`Scenario`, `Input`), so whichever PR merges second rebases without
conflict. If #30873 merges first, these hunks drop out.

**New check.** `scripts/docshtmlcheck` (`make lint/docs-html`, part of
`make lint`) now also reports `empty-table-header`. I put it there
because that checker already parses every Markdown file under `docs/`,
generated pages included, with goldmark and tokenizes raw HTML with
`golang.org/x/net/html`, and it already runs in CI. A markdownlint
custom rule would have needed a new JS rule and HTML parsing there.
- Markdown pipe tables: the header row, via goldmark's GFM table
extension.
- Raw HTML tables: every `<th>`, tracked across HTML blocks, with
entities decoded and whitespace trimmed, so `<th></th>`,
`<th>&nbsp;</th>`, and `<th><br></th>` are all reported. Markdown
between the HTML blocks of an open `<th>` counts as its text. Tables in
code blocks are ignored.
- Output: `docs/x.md:29: empty-table-header: header cell 1 has no text
(table's first header "Auditor")`, plus a summary footer.
- The tokenizer loop moved out of `scanNode` into a shared
`tokenizeSegments` helper so both checks use it; `scanNode`'s behavior
is unchanged.
- The style guide's checks table gets one row; the README for the
checker describes the rule.

## Testing

- Tests first: `TestCheckTables` (now 22 cases) and
`TestRunEmptyTableHeader` failed against a stub, then pass. The existing
`docshtmlcheck` tests pass unchanged.
- `go run ./scripts/docshtmlcheck` on `main` reports 1,561 empty header
cells (116 CLI pages, port-forwarding, and the four #30873 tables); on
this branch it reports none.
- `go test ./scripts/docshtmlcheck/ ./scripts/clidocgen/
./scripts/docgenenv/`, golangci-lint on the changed packages, `make
lint/docs-html lint/docs-manifest lint/docs-gen-filter lint/emdash`,
`pnpm run check-docs`, and Vale on the changed hand-written pages all
pass.
- `make gen` couldn't run to completion locally (no `protoc-gen-go`);
CI's gen check covers it. The CLI pages were regenerated with the repo's
own target.

Review feedback applied in cfd29a2160 (entities decoded in Markdown
text, image alt text and autolinks count as text), 709a58f603 (per-kind
counts, finding order, CLI generator note), 8e61ff3afa (CORS matrix
scopes and caption), and b807bcbbc7 and 2871f89bdf (style guide rule
section, `make lint/docs-html` in the checks table, coverage counts).

Linear: DOCS-1308

---

🤖 Built with AI assistance.
2026-10-21 10:58:04 -07:00
Nick Vigilante d187dea338 docs: end every page description with a period and lint for it (#30863)
## Summary

Page descriptions in the docs were inconsistent about ending with a
period. This makes every page description end with one and adds a lint
so it stays that way.

A page description can come from three places, and the docs site reads
front matter first and falls back to the manifest:

- the hand-written YAML sidebar sources under `docs/manifest/`
(including `children_from` overlays),
- the generated fragments (`docs/manifest/generated/cli.json`,
`rest-api.json`), compiled into `docs/manifest.json`,
- a page's own front matter (today only the generated CLI and API
reference pages carry one).

The check covers all three. For the CLI, the description comes from each
command's `Short`, which stays unpunctuated by CLI-help convention, so
`clidocgen` adds the period when it writes the docs description instead
of changing 120 `Short` strings and the CLI golden files.

## What the lint checks

`make lint/docs-manifest` (`docsmanifestgen check`, already part of
`make lint` in CI) now fails when a non-empty description doesn't end
with `.`. No description in the tree ends with `?` or `!`, so the rule
allows only a period. Errors name the source file, the sidebar
breadcrumb, and the description, for example:

```
install/plan/sizing.yml: Install > Plan your deployment > Coder Validated Architecture > Up to 1,000 users: description must end with a period: "Hardware specifications and architecture guidance for Coder deployments that support up to 1,000 users"
generated/api.json: General: description must end with a period: "General endpoints"
reference/cli/ping.md: Reference > Command Line > `ping`: front matter description must end with a period: "Ping a workspace"
```

The style guide's page-description rule now lists the requirement and
cites the checker, and `styleclaims` recognises
`scripts/docsmanifestgen` so the coverage table counts it as
tool-checked.

## Count of descriptions fixed

147 descriptions in `docs/manifest.json`:

- 27 hand-written descriptions in 15 sidebar source files,
- 120 generated CLI descriptions, which also update the front matter of
120 CLI reference pages.

The REST API descriptions already ended with a period. No description is
otherwise reworded.

## Testing

- New table cases in `scripts/docgenenv` (section route, child route,
overlay, generated fragment, front matter, and a `?` ending) and
`scripts/clidocgen`, written first and failing before the change.
- `docsmanifestgen check` on the unfixed tree reported the 27 source
descriptions; after the fixes and regeneration it passes.
- `go test ./scripts/docgenenv/ ./scripts/clidocgen/
./scripts/docsmanifestgen/ ./scripts/styleclaims/
./scripts/apidocgen/...`
- `golangci-lint run` on the changed packages, `pnpm run check-docs`,
`go run ./scripts/styleclaims`, `scripts/check_emdash.sh`, and
`docsmanifestgen fmt` (no changes).

Linear: [DOCS-1304](https://linear.app/codercom/issue/DOCS-1304)

---

🤖 Built with AI assistance.
2026-10-20 18:51:44 -04:00
Denis AfonsoandClaude 7b8d480ae8 fix(scripts/releaser): use the v-prefixed image tag in release notes (#31032)
## Problem

The release wizard writes this into every release's notes:

```
docker pull ghcr.io/coder/coder:2.37.5
```

but the workflow tags images with `scripts/image_tag.sh`, which always
emits `v<version>`. `ghcr.io/coder/coder:2.37.5` returns 404;
`ghcr.io/coder/coder:v2.37.5` is the real tag. Checked against v2.37.5,
v2.37.4, v2.38.0 and v2.29.19 on ghcr.io: only the `v`-prefixed
manifests exist.

`release.go` strips the `v` before formatting the tag
(`strings.TrimPrefix(newVersion.String(), "v")`). The old bash generator
used `image_tag.sh` directly and got it right.

## Change

Format the tag with `newVersion.String()` (keeps the `v`), with a
comment pointing at `image_tag.sh` as the source of truth.

Affected published notes (36 releases back to v2.29.19) can be fixed
with `gh release edit --notes-file`; this PR only fixes future ones.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Part of ENT-224

Co-authored-by: Claude <noreply@anthropic.com>
2026-10-20 22:02:53 +01:00
Ethan 86ea5d4c4e ci: bump paralleltestctx to detect timeout contexts used through closures (#31039)
#31011 fixed a flake where a closure in `TestProjectSkillQueries` used
the parent test's `testutil.Context` from parallel subtests. Once that
deadline passed, the closure's `require.NoError` called `FailNow` on the
parent `t`, which panics and takes down the whole `coderd/database`
package. `paralleltestctx` only checked direct references to the
context, so it missed this. coder/paralleltestctx#6 now follows calls to
local closures too.

This pins `paralleltestctx` to the current `main` commit, as there's no
tag yet. That commit also brings in the outer variable reassignment
check from coder/paralleltestctx#4 and the `x/tools` bump from
coder/paralleltestctx#5. The closure check flagged the `loadPrompt`
helper in `TestMediaToolResultRoundTrip`, so it now makes its own
context from the subtest's `t`. The `TestGraph` and `TestRuntimeAudit`
fixes for the outer variable check are the same as in #30288, so that PR
only needs to repin.
2026-10-20 15:41:25 +11:00
coder[bot]andDmitry Makovey 116660193e chore: bump bundled terraform to 1.16.6 (#30516)
Bump the bundled Terraform binary from 1.16.2 to 1.16.6 (latest stable,
[released
2026-10-14](https://github.com/hashicorp/terraform/releases/tag/v1.16.6)),
following the same file set as #29331:

- `provisioner/terraform/install.go`: `TerraformVersion`
(`maxTerraformVersion` stays at 1.16.9 since this is a patch bump)
- `install.sh`, `scripts/Dockerfile.base`
- `mise.toml`, `mise.lock` (checksums from HashiCorp's
`terraform_1.16.6_SHA256SUMS`)
- `flake.nix` (fetchurl SRI hashes derived from the same SHA256SUMS)
- `docs/install/prepare/airgap.md`: the example Dockerfile `ARG
TERRAFORM_VERSION` had drifted back to 1.11.0 after the docs move;
updated to 1.16.6

Verified locally with `go test ./provisioner/terraform/ -run
'TestInstall|Version'`.

Opened by Superblink on behalf of @droopy4096.

_Updated from 1.16.5 to 1.16.6 (adds an HTTP/2 resource-exhaustion fix).
Branch name still says 1.16.5. Fixture fingerprint (`generation.sha1`)
regenerated; `go test ./provisioner/terraform/ -run
'TestInstall|Version|Golden'` passes._

---------

Co-authored-by: coder[bot] <137810501+coder[bot]@users.noreply.github.com>
Co-authored-by: Dmitry Makovey <dmakovey@yahoo.com>
2026-10-15 12:33:45 -07:00
coder[bot]andDmitry Makovey 17ec9459c0 fix(scripts/dogfood): pass GO_LINT_NEW_CODE_BASE to the image lint run (#30851)
Fixes the `build_image (26.04)` failure in the `dogfood` workflow on PRs
into `main` (seen on #30807). Tracked in
[ENT-213](https://linear.app/codercom/issue/ENT-213).

## Problem

`lint/go/new-code` (added in #30622) runs `.golangci.new-code.yaml` with
`--new-from-merge-base=$(GO_LINT_NEW_CODE_BASE)`, which defaults to
`origin/main`. `ci.yaml` sets `GO_LINT_NEW_CODE_BASE` to `HEAD^1` for
PRs and `HEAD` otherwise. `scripts/dogfood/test_image.sh` doesn't pass
it into the container, and the dogfood checkout (depth 2) has no
`origin/main`. So `make lint` inside the image reports the whole tree
(1263 issues: errcheck 437, forbidigo 42, nolintlint 784) and fails
"Test image tooling".

Pushes to `main` run the same validation before pushing the image, so
the next dogfood-triggering merge to `main` would also fail and skip the
Docker Hub push.

## Change

`scripts/dogfood/test_image.sh` only:

- Under GitHub Actions, and when `GO_LINT_NEW_CODE_BASE` isn't already
set, default it to `HEAD^1` for `pull_request` and `HEAD` otherwise,
mirroring `ci.yaml`.
- Forward it into the container with `--env GO_LINT_NEW_CODE_BASE`.
- Local runs are unchanged: the variable stays unset unless you set it,
so the Makefile default (`origin/main`) applies.

This PR touches `scripts/dogfood/**`, which triggers the `dogfood`
workflow, so its own `build_image (26.04)` run is the end-to-end test.

Opened by Superblink on behalf of @droopy4096.

Co-authored-by: coder[bot] <137810501+coder[bot]@users.noreply.github.com>
Co-authored-by: Dmitry Makovey <dmakovey@yahoo.com>
2026-10-15 11:11:10 -07:00
Bobby Ho 603dd420d1 feat: add ServerTailnet logs and metrics for unreachable agents (#30133)
Add default-level metrics and warn-level diagnostics to investigate
ServerTailnet reachability failures causing agent-specific 502s, without
restarting coderd or enabling debug logging.

- Enrich request logs and PTY/chatd warnings with agent ID, failure
  reason, wait duration, coordinator state, WireGuard handshake and
  byte counts, and DERP regions.
- Add metrics for unreachable attempts by reason, reachability latency,
  embedded DERP connections, and offline app/terminal requests.
- Distinguish reachability failures from subsequent TCP dial failures,
  cap reverse proxy dials at one minute, and exclude expected chatd
  readiness polls and superseded-agent dials from unreachable counts.
- Report missing handshakes as zero time instead of 1970, also fixing
  `coder ping`.

Part of PLAT-680, a sub-issue of PLAT-360.
2026-10-14 14:55:29 -07:00
Michael Suchacz e7c2fd5d2e feat: add Gemini 4 Argon to Agents and AI Gateway (#30780)
Add Gemini 4 Argon to the Agents known models catalog, so it appears as
a suggestion when an admin adds a chat model, and regenerate the AI
Gateway price book from models.dev.

| Provider | Model ID | Placed before |
|---|---|---|
| google | `gemini-4-argon` | `gemini-3.1-pro-preview` |
| openrouter, vercel | `google/gemini-4-argon` |
`google/gemini-3.8-flash` |

**Pricing.** The price book carries Google's launch rates on google,
openrouter, vercel, and copilot: $2 per 1M input tokens, $10 per 1M
output tokens, and $0.10 per 1M cached input tokens. Google doubles them
on 2026-12-01 ($4, $20, $0.20); the weekly refresh picks that up once
models.dev changes. The google entry keeps models.dev's 32,768 max
output tokens, the cap Google enforces on non-streaming
`generateContent` calls (larger values return HTTP 400 there).

**Thinking levels.** `googleSupportedThinkingLevels` did not recognize
the `argon` name token and failed closed, so chatd sent Argon no
`thinking_level` and no `include_thoughts`: the native Google path
dropped the reasoning effort, and neither path showed thought summaries.
Argon accepts LOW, MEDIUM, and HIGH like Gemini 3.1 Pro and rejects
MINIMAL, so the map now treats it like the Pro family.

**Upstream drift.** `prices.json` covers every model of its 8 providers,
so the regeneration also brings the changes since the last weekly
refresh, all on OpenRouter and Vercel: 14 other new entries, 26 entries
models.dev delisted, and 51 price changes. The catalog picks up the same
changes for 8 curated OpenRouter entries (DeepSeek, Kimi, Qwen, and
GLM).

## Testing

- Live requests to `gemini-4-argon` through the AI Gateway's Google
provider (Google's OpenAI-compatible endpoint): `reasoning_effort` and
`thinking_level` values `low`, `medium`, and `high` return 200 and
`minimal` returns 400; `include_thoughts` returns thought text in
streaming and non-streaming responses.
- Red-green: removing the `argon` token fails
`TestGoogleSupportedThinkingLevels/gemini-4-argon` (expected LOW,
MEDIUM, HIGH; got nil).
- Regenerating from the saved models.dev snapshot reproduces both
generated files byte for byte.
- `go test ./scripts/aibridgepricesgen/... ./coderd/aibridge/prices/...
./coderd/x/chatd/chatprovider/...` and `pnpm exec vitest run
--project=unit src/modules/aiModels/` pass. The pre-commit hook passed;
local pre-push checks were skipped, so CI is the first full run.

## Review record

- Frontend review: the only `site/` change is the generated
`knownModelsGenerated.json`, so the FE1 to FE10 checklist does not
apply; no hand-written frontend code changed.

> Xum opened this PR on behalf of @ibetitsmike.
2026-10-14 22:39:00 +02:00
Denis AfonsoandClaude 200cffcfd8 fix(scripts/release): promote stable on immutable releases (#30745)
## Problem

`./scripts/release_promote_stable.sh v2.37.4` failed today with:

```
edit release failed: PATCH https://api.github.com/repos/coder/coder/releases/403906567: 404 Not Found
```

Releases on this repo are immutable (`gh release view v2.37.4 --json
isImmutable` → `true`). `promoteVersionToStable` cloned the entire
`RepositoryRelease` into `EditRelease`, so the PATCH carried `tag_name`,
`target_commitish` and the assets. GitHub rejects that on an immutable
release even when the values are unchanged, and answers 404 rather than
422. The same edit with only `body`, `prerelease` and `make_latest`
succeeds (that's how v2.37.4 was promoted by hand).

## Change

- Build a minimal `RepositoryRelease{Body, Prerelease, MakeLatest}` for
the edit.
- Drop the `Draft: false` assignment (published releases are never
drafts, and `draft` is not an editable field on immutable releases).
- Dry-run now logs the body diff instead of a diff of the whole object.
- Remove the now-unused `cloneRelease` helper.

Verified with `go test ./scripts/release/` and a `promote --dry-run`
against v2.36.7.

Every release published since immutable releases were enabled is
affected, so without this the command cannot promote anything. Related:
the release-process review notes that this step is also missing from the
runbook and should become a workflow job on every `.0` mainline release.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-authored-by: Claude <noreply@anthropic.com>
2026-10-14 16:42:56 +01:00
Michael Suchacz 7517301ffa chore: add golangci new-code quality gates (#30622)
Stack context: second layer of the Go lint gate stack, built on the
exclusion fix below.

Why: adds commented-out code detection and a `log/slog` ban to the main
config, plus a new-code-only pass requiring reasons on lint
suppressions, flagging ignored errors (including `_, _ = w.Write(...)`),
and banning `time.Sleep` in tests. Fixing the hits surfaced a real bug:
GitLab warnings went to the wrong logger. The new-code pass resolves
symlinked checkout paths so it does not silently report nothing.

Validation: both configs report 0 issues; each seeded violation is
caught; `make pre-commit` passed.

> Xum acted for Mike to open this PR.
2026-10-13 07:57:56 +00:00
Michael Suchacz e8dcf9a3f1 fix: anchor golangci .git exclusion and fix hidden findings (#30621)
Stack context: two-layer stack tightening Go lint gates. This bottom
layer fixes the lint exclusion; the layer above adds new rules.

Why: a `.git` path exclusion in the golangci config was hiding Go files,
masking 70 findings. This fixes the exclusion, fixes the findings
(`agent/agentgit` now runs git through `agentexec.Execer`), and adds a
`make lint/go` check that fails when an exclusion hides Go files. `make
lint/go` now waits for a concurrent golangci-lint run instead of
failing. The exhaustruct exclusion rule for `scripts/*` had the same
problem: as a regex it also matched `agent/agentscripts` and
`coderd/agentapi/scripts.go`, so it is now `^scripts/`.

Validation: full-tree lint reports 0 issues; the new check fails on the
old entry and passes on the new one.

> Xum acted for Mike to open and update this PR.
2026-10-13 07:57:56 +00:00
Ehab Younes 2f2169ed19 fix: deduplicate template insights family metrics (#30497)
coderd_insights_applications_usage_seconds summed each family's per-app
minutes, so a user with VS Code and Cursor open in the same minute was
counted as two VS Code minutes. GetTemplateInsightsByTemplate now takes
the registry's app-to-family mapping as jsonb, groups the raw agent
stats by family, and counts each minute once per user and family.

- Keep the metric's four built-in series (Visual Studio Code, JetBrains,
  Web Terminal, SSH) and their labels; only their values change. The
  HELP text now says built-in apps are grouped by family.
- Add AppFamilyName.DisplayName() and remove the duplicate
  TemplateBuiltinAppDisplayName* constants from codersdk. The template
  insights handler returns the same display names.
- Pin how the migration 000596 rollback approximates family minutes
  from per-app rows.

Family values can drop wherever the old per-app sum double-counted
overlapping apps. No migration or storage changes.
2026-10-12 23:58:35 +03:00
Ehab Younes 5b060beeb1 feat: separate connection log methods from app names (#30466)
Connection logs now record how a connection was made
(`connection_method`) separately from which app made it (`app_name`).
Storage, the HTTP API, search filters, and the UI move to the new
fields. Legacy agent reports and the deprecated `type` field keep
working.

- Add `connection_method` (`ssh`, `reconnecting_pty`, `workspace_app`,
  `port_forwarding`, `tunnel`) and optional `app_name` and
  `app_display_name` to `GET /api/v2/connectionlog`. Deprecate `type`
  and make it a plain string in the API and the Go SDK.
- Add `method:` and repeatable `app:` search filters. `type:` keeps its
  old results but can't be combined with the new filters.
- Replace the Type menu with a Method menu, show the app next to the
  method in each row, and warn when a filter uses `type:`.
- Change the Active SSH connections preset to `method:ssh`, so it also
  includes SSH sessions from VS Code and JetBrains apps.
- Generate `codersdk.ConnectionLogMethod` from the database enum with
  `typegen dbenums`.
- Migration 000617 rewrites `connection_logs` under an ACCESS EXCLUSIVE
  lock, which takes about 50 seconds for 10 million rows. `vscode` and
  `jetbrains` rows become `ssh` rows with that app. Rolling back folds
  apps into the old types and loses per-app names.
- Document the new filters and how to replace each `type:` value.

Closes #27412
2026-10-12 15:56:40 +03:00
Thomas Kosiewski f923f4a525 chore!: remove the chat automations experiment (#30412)
Removes the experimental Chat Automations feature (`chat-automations` experiment) from main. This reverts the 16-PR stack that added it and adds two forward migrations that undo its schema and credential changes.

> [!WARNING]
> **Upgrade consequences**
>
> - **Automation data is deleted.** Migration 000615 drops the `chat_automations` table and deletes every queued message an automation delivered. Chats, their messages and ordinary queued messages stay. Messages lose their automation provenance (`automation_id`, `input_id`).
> - **Automation-only credentials are revoked.** Migration 000616 deletes API keys, OAuth2 codes and OAuth2 tokens whose only scopes (or allow-list entries) were `chat_automation:*`. Mixed credentials keep their other scopes. Custom roles lose their `chat_automation` permissions. OAuth2 apps that allowed only `chat_automation` scopes get the whitespace allowlist that grants nothing, as in 000595.
> - **A coordinated restart is required.** Stop all old replicas before the new version runs the migrations. Old replicas read the dropped table and columns, so a rolling deploy is not supported.
> - **Remove `--chat-max-automations-per-owner` before upgrading.** The flag, the YAML key `chat.maxAutomationsPerOwner` and the env var `CODER_CHAT_MAX_AUTOMATIONS_PER_OWNER` are removed. A deployment that still sets the flag or the YAML key will not start (`unknown flag` / `unknown option`). The env var is ignored. This includes dogfood. No release or tag contains the flag.
> - **Down migrations restore the schema only.** 000616 down and 000615 down recreate the 000611 schema with empty tables. They do not restore deleted automations, webhook secrets, provenance, queued deliveries, revoked credentials, stripped permissions or the deleted experiment rule.

## What is removed

- The `chat-automations` experiment and its stored runtime rule (`experiment_rule:chat-automations`).
- The `--chat-max-automations-per-owner` deployment option.
- The `chat_automation` RBAC resource, its member and chatd grants, and the `chat_automation:*` API key scopes.
- The automation sqlc queries, dbauthz wrappers and audit table entries.
- The management routes under `/api/experimental/organizations/{organization}/chat-automations`, and the webhook route `POST /api/experimental/chat-automations/{automation}/events` with its rate limiter.
- The chatd management service, webhook publish, schedule loop, the `manage_automations` tool, and the chatstate admission seam and queue promotion guard that only automations used.
- The Agents Automations page, editor, nav entry, chat labels, the Manage automations toggle and the API client.
- The `archived:any` chat search value that the automations "View chats" dialog added.
- `docs/ai-coder/agents/automations.md`, its manifest entry, the glossary entry and the related rows.

Reverted PRs: #30085, #30096, #30112, #30117, #30127, #30134, #30160, #30166, #30167, #30171, #30187, #30207, #30214, #30217, #30219, #30220.

## What stays, and why

- Migrations 000609, 000610 and 000611, their fixture, and the #30297 lock fix with `migration000609_test.go` stay unchanged. Deployments such as dogfood already ran them.
- New migration **000615** reverses the schema changes of 000609 and 000611 against the current schema. It uses the same short-wait lock-and-retry block as 000609, and it rebuilds `chats_expanded` from main's current definition minus the two automation columns. New migration **000616** follows the 000595 Tasks removal for credentials and recreates `api_key_scope` without the five `chat_automation:*` values. It keeps the `chat_project_memory:*` values from 000612.
- The `chat_automation` value stays in the `resource_type` enum, and `codersdk.ResourceTypeChatAutomation` stays. Old audit log rows stay readable, as with `task` after the Tasks removal.
- The 000615 delete fires the existing queue-version trigger, so `queue_version` moves to `snapshot_version` for each affected chat. `snapshot_version` is not bumped: the coordinated restart restarts every stream loop from an empty state, and the site does not read the version fields.
- `coderd/x/chatd/ARCHITECTURE.md` drops the sections the stack added and gets back its pre-stack wording, merged with the later main edits to the same file.

## Notes for reviewers

- #30379 (chat project page) merged while this PR was in progress. It passed `manageAutomationsEnabled` in `AgentCreatePage.test.tsx` and listed an `automations` route in `AgentsPageLayout.test.tsx`. Later main commits also added automation props in `ConversationTimeline.test.tsx` and `ConversationTimelineWorkingBlocks.stories.tsx`. This PR removes all of them.
- Remaining `automation` matches under `coderd`, `codersdk`, `enterprise`, `site/src` and `docs` are intended:
  - migrations 000609 to 000611, their fixture and `migration000609_test.go`;
  - the new 000615 and 000616 migrations and `migration000615_test.go`;
  - the `chat_automation` resource type in `dump.sql`, `models.go`, `codersdk/audit.go`, the generated API docs, `swagger.json` and `typesGenerated.ts`;
  - unrelated uses of the word "automation" (docs, `chatlabels_test.go`, `ratelimit.go`, `autostop.go`, `subagent_catalog.go`, `WorkspaceSharingSection.tsx`, `AgentChatPage.stories.tsx`).

Remote UAT on an isolated dev deployment passed. It upgraded a populated pre-removal database, then checked chats, queued messages, tokens, the 404 routes, the removed page and nav entry, and the experiments list.

---
_Generated with [`xum`](https://github.com/coder/xum) • Model: `anthropic:claude-opus-5-5` • Thinking: `high`_

![Agents sidebar after the upgrade, with no Automations entry](https://github.com/user-attachments/assets/bd446035-7fc8-4ca4-a735-3a4f36380738)
2026-10-11 19:57:49 +02:00
Nick Vigilante 35b3052ea8 ci(scripts/docsredirectscheck): reject redirects an earlier rule swallows (#30383)
## Summary

Redirects apply in file order and the first match wins, so rules have to
run from most specific to least.
Nothing enforced that, so a wildcard placed above a narrower rule inside
its subtree silently swallowed it.

## Problem

Given these two rules in that order, the second never applies:

```json
[
  { "source": "/docs/retired/:section(.*)", "destination": "/docs/reference/:section" },
  { "source": "/docs/retired/foo",          "destination": "/docs/about" }
]
```

A reader asking for `/docs/retired/foo` is sent to `/docs/reference/foo`
instead of `/docs/about`.
If `/docs/reference/foo` happens to exist, every existing check passes:
the destination resolves, there is no duplicate source, no chain, and no
loop.
The rule is simply dead, and the only symptom is readers landing on the
wrong page.

`docs/redirects.json` is empty today, so this adds the guard before the
file gets its first wildcard rather than after.

## Fix

Compare each rule against the ones before it and fail when an earlier
source matches every URL a later source would.

Coverage is decided on path boundaries, not string prefixes, so
`/docs/retired/:path*` does not cover `/docs/retired-too/page`.
Two cases stay legal on purpose:

- An exact rule above a wildcard on the same prefix, which claims the
bare prefix and leaves the subtree to the wildcard.
- A regex wildcard above an exact rule on its bare prefix, since
`/:name(.*)` never matches the prefix itself.

Rules that share a source are still reported as duplicates, not as
unreachable, so one mistake produces one message.

## Changes

- `rules.go`: add `pattern.covers` and `checkOrder`, called from
`checkAcrossRules`.
- `README.md`: document the ordering requirement and the new failure.
- `main_test.go`: 11 cases covering each pattern pairing, both
orderings, the path-boundary case, reporting once against the first
coverer, and the duplicate interaction.

## Testing

- `go test ./scripts/docsredirectscheck/` passes;
`TestCheckRedirectsOrder` runs all 11 subtests.
- `make lint/docs-redirects` passes against the committed
`docs/redirects.json`.
- `gofmt -l` and `go vet` are clean on the package.
- Ran the binary against the example above: it fails with
`unreachable: rule [0] (source "/docs/retired/:section(.*)") comes first
and matches every URL this rule would; order rules from most specific to
least`,
  exits 1, and passes with exit 0 once the two rules are swapped.

<details>
<summary>Why this check, and what it is a prerequisite for</summary>

`docs/redirects.json` is validated in CI and triggers a docs rebuild,
but no site reads it yet.
The redirects actually served live in two other hand-maintained files:
`coder.com/redirects.json` (297 rules, 173 under `/docs`) and
`docs-engine`'s `apps/router/redirects.json` (221 rules).

The plan is to make `docs/redirects.json` the single authored source and
have docs-engine compile it per version, so `apps/router/redirects.json`
becomes generated output.
Classifying the router's 221 rules showed all of them are same-version
moves once the version prefix is stripped, and that expanding them into
this file's narrow grammar yields 279 authored rules.

That expansion is what makes this check necessary now.
Collapsing many exact rules into wildcards is an engine optimization,
and the authored file will carry both wildcards and exact rules in the
same subtree, which is exactly the shape that can silently swallow a
rule.
Adding the guard while the file is still empty means the migration
cannot introduce an unreachable rule without CI saying so.

</details>

---

Resolves DOCS-1188.

Opened by Coder Agents on behalf of @nickvigilante.

<details>
<summary>CI note: one unrelated flake</summary>

`test-go-pg (ubuntu-latest)` failed on `coderd/x/chatd
TestChatTurnStopsAtStepLimit/Default`.
This PR changes three files under `scripts/docsredirectscheck` and
nothing in `coderd/`, so the failure is unrelated to the diff.
`lint`, `gen`, `fmt`, `check-build`, `test-go-race-pg`, `test-go-pg-17`,
`test-e2e`, and `test-go-pg` on macOS and Windows all pass.

The first run of this PR also had `changes`, `cla`, and `release-labels`
cancelled rather than failed, all three at exactly 15m03s, which skipped
the Go jobs entirely. A `--failed` rerun produced the results above.

</details>
2026-10-08 12:26:20 -04:00
Nick Vigilante bcedc5d47e fix(scripts/docsgenfiltercheck): accept CRLF line endings in banner check (#30476)
`bodyOpensWithContentBanner` split pages on `\n` and compared lines
exactly.
On a Windows checkout with CRLF endings, the first line of
`docs/reference/cli/index.md` is `---\r`, so the front matter was never
recognized and `TestBodyOpensWithContentBanner` failed deterministically
on `test-go-pg (windows-2022)`.

Normalize CRLF to LF before parsing, and add CRLF table cases so the bug
reproduces on Linux too.

Fixes CLI-55.
Failing run: https://github.com/coder/coder/actions/runs/37570448753

## Testing

- The new CRLF cases failed before the fix and pass after it.
- `go test ./scripts/docsgenfiltercheck`, `GOOS=windows go vet`,
`golangci-lint`, and the full pre-commit hook pass.

<details>
<summary>CI flake note</summary>

The first `fmt` run failed while downloading Go modules
(`proxy.golang.org` returned `stream error: ... INTERNAL_ERROR` for
`testcontainers-go@v0.40.0`), which also failed `required`. This is a
network error unrelated to the change. I re-ran the failed jobs.

</details>

---

_This PR was generated by Coder Agents on behalf of @nickvigilante._
2026-10-08 12:24:42 -04:00
Nick Vigilante d031680ff2 fix(scripts/metricsdocgen/scanner): normalize package keys to slash paths (#30360)
Fixes the Windows failures in `TestBuildPrefixIndex` and `TestExcluded`
(ENG-3443). The scanner keyed package directories with `filepath.Dir`,
which uses `\` on Windows, but compared them against keys derived from
Go import paths and against `excludeDirs`, which always use `/`.
Prefixes never reached imported packages, and excluded subtrees weren't
skipped.

Each parsed file now records its directory once in slash form
(`packageDir`), and the `prefixIndex` doc states the key rule.
`excluded` normalizes its input because `scanDirectory` passes it native
`filepath.WalkDir` paths. Generated output on Linux is unchanged.

PR CI only runs `./cli/... ./enterprise/cli/... ./agent/...` on Windows,
so I dispatched [`nightly-gauntlet` on this
branch](https://github.com/coder/coder/actions/runs/37316261411). On
`windows-2022`, both `scripts/metricsdocgen` and
`scripts/metricsdocgen/scanner` pass. That run tested `c5462169cd`. The
only later commit drops an unused field and rewords comments, so the
result still applies.

<details>
<summary>Unrelated failures in the gauntlet run</summary>

The Windows job still fails on `TestWorkspaceToolsGolden`,
`TestEditMessageCancelsToolCallsOnAgent`, and
`TestTools/WorkspaceExecute`. These fail the same way in [main's
2026-10-05
nightly](https://github.com/coder/coder/actions/runs/37262889155),
together with the scanner failures this PR fixes.
`TestAcquirer_MatchTags` also failed once, in provisioner code that this
PR doesn't touch.

</details>

---

🤖 Generated by Coder Agents on behalf of @nickvigilante.
2026-10-06 14:53:20 -04:00
Nick Vigilante 28cf704f27 chore: update references to the relocated install docs paths (#30380)
The Install docs moved into the five-phase spine in #29304, but several
references outside `docs/` still point at the pre-move URLs.
`docs/redirects.json` is empty, so those URLs 404 rather than
forwarding.
The release note generators are the user-visible ones: every generated
release note links `coder.com/docs/install/releases` and
`coder.com/docs/install/upgrade`.

| Reference | Was | Now |
|---|---|---|
| Release schedule links in release notes | `/docs/install/releases` |
`/docs/reference/releases` |
| Upgrade link in release notes | `/docs/install/upgrade` |
`/docs/install/operate/upgrade` |
| README sizing link |
`/docs/admin/infrastructure/validated-architectures` |
`/docs/install/plan/sizing` |
| Provisioner tag table comments | `docs/admin/provisioners/index.md` |
`docs/install/operate/provisioners/index.md` |
| FAQ provenance comment | `docs/admin/licensing/index.md` |
`docs/install/prepare/licensing.md` |
| ESR 2.24-2.29 upgrade table | `/docs/admin/provisioners` |
`/docs/install/operate/provisioners` |
| Helm provisioner README, `NOTES.txt`, and `values.yaml` |
`/docs/admin/provisioners` | `/docs/install/operate/provisioners` |
| Helm Coder chart README | `/docs/install/kubernetes` |
`/docs/install/server/kubernetes` |
| `workspace-tags` example README |
`/docs/admin/provisioners#external-provisioners` |
`/docs/install/operate/provisioners#types-of-provisioners` |
| `kubernetes-envbox` example README |
`/docs/admin/infrastructure/architecture#agents` |
`/docs/install/plan/architecture#agents` |

`scripts/releaser/docs.go` was already updated in #29304; this covers
what that pass missed.

Two notes on anchors:

- The `#external-provisioners` anchor was already dead before #29304, so
it now points at `#types-of-provisioners`, the section covering
external, organization-scoped, and user-scoped provisioners.
- Every other anchor was verified to still exist on its new page.

One link left alone: `provisionerd/provisionerd.go` references
`docs/admin/provisioners.md` through a pinned-commit GitHub permalink,
which still resolves and shouldn't be rewritten.

## Verification

- `go test ./scripts/release/...` passes, including the release note
golden strings.
- `go build ./scripts/releaser/` succeeds.
- `make pre-commit` passes.
- Markdown table padding reflowed with `markdown-table-formatter`;
`markdownlint` reports no new findings.

<sub>Opened by Coder Agents on behalf of @nickvigilante.</sub>
2026-10-05 22:28:54 +00:00
Nick Vigilante b9c0853473 ci: skip the gen job for docs-only PRs that cannot change its output (#30164)
The CI `gen` job ran `make -j -B gen` on every PR, so a docs-only PR
spent about 160 seconds of an 8-core runner (12 recent runs: 154 to 170
seconds) confirming that nothing changed.
This PR skips the job for docs-only PRs that touch none of the docs
paths `make gen` reads or rewrites.
Replaying every commit on `main` since 2026-08-01 through the new
filter, 96 of 116 docs-only commits (83%) would skip.

- A new `docs-gen` paths filter lists `docs/manifest.json`,
`docs/reference/api/**`, `docs/reference/cli/**`, and the three
generated admin pages.
`gen` now needs `changes` and runs unless the PR is docs-only with no
`docs-gen` or workflow match.
Push, release-branch `create`, and `workflow_dispatch` runs always
regenerate. Non-release `create` events now skip `gen`, like every other
job that needs `changes`.
- `lint/docs-gen-filter` (`scripts/docsgenfiltercheck`, run by `make
lint`) treats as generated every file in `GEN_FILES`, plus every tracked
Markdown page whose body opens with the shared
`docgenenv.GeneratedContentBanner`. It fails when a generated file
matches the `docs` filter but not `docs-gen`, or when a `docs-gen`
pattern matches no tracked file.
  This keeps the filter from drifting when someone adds a generator.

## Why skipping is safe

- No generator reads docs prose. Makefile rules and generators only
touch `docs/manifest.json` and the generated pages under `docs/`;
`examplegen` embeds only `examples/templates`, and no golden-file test
reads `docs/`.
- `dorny/paths-filter` matches a renamed file on both its old and new
paths (`src/main.ts` at the pinned SHA), so moving a generated page out
of `docs/reference/cli/` still runs `gen`.
- If `changes` fails, `gen` is skipped and `required` fails on the
`changes` result, so a failed classification never counts as an
exemption.

Docs-only PRs that edit `docs/manifest.json` (usually a new page) or a
generated page still run the full job, because the generators rewrite
those files.
DOCS-1142 tracks a lighter path for those PRs and for fresh-worktree
rebuilds: render the docs from the committed `swagger.json` and sidebar
fragments instead of running swag or building the CLI.
That work depends on DOCS-1131, which rewrites the same Makefile rules.

Refs DOCS-1141

<details>
<summary>Measurements, options considered, and validation</summary>

**Local profile (not changed by this PR).**
In a warm worktree, editing a docs page or `manifest.json` and running
`make -j gen` rebuilds nothing (0.9 seconds).
In a fresh worktree with a warm Go cache on 12 cores, `make -j
docs/manifest.json` took 40 seconds and `make -j gen` took 46 seconds.
The time went to `_gen/bin/clidocgen` (26 seconds), swag in
`coderd/apidoc/.gen` (18 seconds), and the database codegen chain, which
needs PostgreSQL.
Checkout mtimes are arbitrary and `coderd/apidoc/.gen` is untracked, so
all of that reruns once.
`clidocgen` runs twice per `make gen`, but each run takes about 1
second, so merging the runs isn't worth it.

**Options considered.**

1. Skip `gen` by path for docs-only PRs (this PR). Low risk, independent
of DOCS-1131.
2. A light CI job that runs only the docs generators, skipping heavy
prerequisites with `make -o`. Rejected: CI would keep a second
dependency list that can drift from the Makefile.
3. Content-hash stamps for swag. A committed stamp would conflict on
nearly every backend PR, and an untracked one doesn't help fresh
worktrees. Deferred to DOCS-1142.
4. Split swag from the API Markdown rendering now. It conflicts with
DOCS-1131. Deferred to DOCS-1142.

**Filter validation** (the workflow's own filters with `picomatch@2`,
`dot: true`, the matcher `paths-filter` uses):

| Change | `gen` |
| --- | --- |
| Prose page, image, root `README.md`, style guide, `.claude/docs` |
skips |
| New page plus `docs/manifest.json` | runs |
| Edit, delete, or rename out of `docs/reference/cli/**` or
`docs/reference/api/**` | runs |
| Audit log, Prometheus, or configuration reference page | runs |
| Mixed code and docs, `Makefile`, generator script, `offlinedocs` |
runs |
| Workflow change | runs |
| Docs-only `push`, release `create`, `workflow_dispatch` | runs |

Running `make lint/docs-gen-filter` against edited copies of the
workflow fails, as it should, in each of these cases:
`docs/reference/cli/**` narrowed to `docs/reference/cli/*.md` (137
nested pages), `docs/reference/api/**` narrowed to `*.json` (31 pages
plus the stale pattern), the configuration reference entry removed, a
mistyped admin page path, and the unsupported patterns
`docs/reference/**.md` and `docs/reference/*/**`.

</details>

<details>
<summary>CI flake unrelated to this PR</summary>

`test-js` failed once, on the second push, in
`src/pages/AISettingsPage/SpendPage/SpendPage.test.tsx` ("applies and
clears the 'provider_name' filter and resets pagination"). The same test
failed on `main` in run 36649227804. This PR changes nothing under
`site/`, and `test-js` passed on every later push.

</details>

---

_Generated by Coder Agents on behalf of @nickvigilante._


<details>
<summary>CI flake note</summary>

`test-go-pg (ubuntu-latest)` failed once on `ff25eba` because the runner
ran out of disk (`No space left on device`) before the tests reported
results.
This PR doesn't change Go code under test. The rerun passed.

</details>
2026-10-05 09:27:46 -04:00
Nick Vigilante fcbf45c571 docs: render audit log resources as one table per resource (#29797)
## Summary

The audit logs page now lists each tracked resource in its own section
with a simple two-column table, replacing the nested HTML tables that
rendered poorly and overflowed on phones.

## Before and after

Before, the generated block was a single Markdown table with a malformed
`|<b>Resource</b>||` header.
Each row held the resource name and its actions in one cell, then split
a raw HTML `<table>` of fields across the remaining cells.
The result rendered unevenly across renderers, was far wider than the
page on mobile, and was hard to navigate with a screen reader.

After, each resource gets an `###` heading under **Tracked Events**,
which also gives it a linkable anchor.
Below the heading, an `Actions:` line lists the audited actions as
inline code, followed by a two-column Field and Tracked table with `Yes`
or `No` values.
Every table uses the same fixed column widths (Field 75%, Tracked 25%),
so the Tracked column lines up across all resources.
Two narrow columns fit at phone width without horizontal scrolling.
Resources with no mapped actions skip the `Actions:` line instead of
rendering an empty one.
Resource ordering is unchanged.

## Changes

- `scripts/auditdocgen/main.go` writes one section per resource.
Each fields table is a small HTML table with `width` attributes, because
Markdown tables size columns to their content and cannot align across
tables.
GitHub and the docs site sanitizer both keep `width` but strip inline
`style` and `<colgroup>`.
- The generator no longer panics when its start marker is the last thing
in the file.
- New `scripts/auditdocgen/main_test.go` covers section output, marker
handling, and idempotent regeneration.
- `docs/admin/security/audit-logs.md` is regenerated, and the intro
sentence above the block now explains the actions line and the Tracked
column.

## How it was regenerated

`make docs/admin/security/audit-logs.md`, which builds the generator and
runs `markdownlint-cli2 --fix` and `markdown-table-formatter` over the
output.
Running it a second time produces no diff.

## Testing

- `go test ./scripts/auditdocgen/...` passes.
- `pnpm exec markdownlint-cli2 docs/admin/security/audit-logs.md`
reports 0 errors.

Linear: https://linear.app/codercom/issue/DOCS-894

---
🤖 Built with AI assistance.
2026-10-02 15:05:10 -04:00
Nick Vigilante 12caf883fc test(scripts/docgenenv): match missing include by fs.ErrNotExist (#30285)
## Summary

`TestBuildManifestErrors/IncludeMissingFile` fails deterministically on
Windows because it matches the POSIX error text `no such file or
directory`.
Windows reports the same condition as `The system cannot find the file
specified.`
This asserts `fs.ErrNotExist` in the error chain instead, plus the
included file name.

Fixes [ENG-3439](https://linear.app/codercom/issue/ENG-3439).

## Changes

- Add an optional `wantErrIs` field to the `TestBuildManifestErrors`
table, checked with `require.ErrorIs`.
- `IncludeMissingFile` now expects `fs.ErrNotExist` and the text
`nope.yml` (not `ref/nope.yml`, since Windows renders the path with
escaped backslashes).

No production code changes: `DecodeSourceFile` already wraps the
`os.ReadFile` error with `%w`, and `errors.Join` preserves it.

## Testing

- `go test ./scripts/docgenenv/... -count=1` passes on Linux.
- `GOOS=windows go test -c ./scripts/docgenenv` compiles.
- `golangci-lint run ./scripts/docgenenv/...` reports 0 issues;
pre-commit hook passed.
- Not run on Windows locally; the `test-go-pg (windows-2022)` job is the
real check.

---

_Generated by Coder Agents on behalf of @nickvigilante._
2026-10-02 14:52:57 -04:00
Nick Vigilante bb58916e6c ci: validate docs/redirects.json (#30209)
Adds `docsredirectscheck`, a lint for `docs/redirects.json`, the list of
redirects the documentation website applies for pages that have moved or
been removed. It runs as `make lint/docs-redirects`, part of `make lint`
and `make lint-light`, so a bad redirect is caught in the pull request
that adds it. The file does not exist yet, and the check passes while it
is absent, so this can merge before or after the file does.

## What it checks

It fails on:

- a file that is not a JSON array of well-formed rules (`source`,
`destination`, optional boolean `permanent`, no unknown fields);
- syntax outside the supported subset: a source under `/docs/`, written
as an exact path, a trailing `/:name*`, or a trailing `/:name(.*)`, with
no version segment such as `/@main`;
- an internal destination that is not a page in `docs/manifest.json`, or
a pattern destination that points into a section with no pages or uses a
parameter the source does not define;
- a source that matches a live page, which would hide it;
- duplicate sources, chains (a destination that is another rule's
source), and loops.

It warns, and never fails, when a pull request removes routes from
`docs/manifest.json` while `docs/redirects.json` is unchanged and does
not already cover them. Pages are sometimes removed on purpose. The
comparison is by route rather than by file, so a rename that keeps its
URL (for example `admin/foo/index.md` to `admin/foo.md`) does not warn.
In GitHub Actions the warning is an annotation on the manifest. The base
is the first parent of the PR merge commit, the same approach
`scripts/check_emdash.sh` uses, and the warning is skipped with a note
when it is not available.

Routes are derived from `docs/manifest.json` the way the docs website
derives them, including its plain-suffix stripping of `README.md`,
`index.md`, and `.md`, so the check and the site agree about which pages
exist. `scripts/docsredirectscheck/README.md` documents the rule format
and the limits.

## Testing

- 128 Go tests pass under Go 1.26.5: route derivation, every failing and
passing rule shape, the absent-file pass, exit codes, and the warning
logic through a fake git reader.
- I checked that the tests can fail by breaking the code 16 ways (for
example dropping the `index.md` suffix, disabling the shadow check,
reporting loops twice). All 16 were caught. That run found one gap, a
missing test for "redirects file changed", which is now covered.
- Ran against the repository's real `docs/manifest.json`: passes with no
redirects file, and a file with one of each failure class reports every
problem with the rule index, the source, and how to fix it.
- Ran the warning against real git history: it warns for an older base,
stays silent at `HEAD`, and skips with a note for an unknown revision.
- Lint: `golangci-lint` v2.0.2 built with Go 1.26.5 against the
repository config reports 0 issues. `paralleltestctx`, `intxcheck`,
`typos` 1.46.1, `gofumpt` and `gci` (through `make fmt/go`), and `go
vet` pass.

CI: `lint` (which runs this check), `fmt`, `title`, and the Go and
end-to-end test jobs pass. The `gen` job, and so `required`, fails on
this branch for a reason this PR does not touch: `main` is failing it
too. `docs/manifest/generated/cli.json` has not been regenerated for the
`coder sharing` commands that landed in the latest commit on `main`. It
should clear once `main` is fixed and merged into this branch.

Not run locally: the full `make lint` and `make pre-commit`. The local
hooks are opt-in and were off. The annotation output has been exercised
by unit tests and a local run with `GITHUB_ACTIONS=true`, but not yet on
a real pull request merge commit; this PR does not change the manifest,
so it will not exercise it either. No workflow change is needed, because
the `lint` job already runs `make lint` for any change that is not
offlinedocs-only.

---
🤖 Built with AI assistance.
2026-10-02 14:27:44 -04:00
Thomas Kosiewski 9316d67440 feat: add chat automations management API behind chat-automations (#30112)
Adds the management API for chat automations under
`/api/experimental/organizations/{organization}/chat-automations`,
behind the new user-scoped `chat-automations` experiment. This is the
first of two PRs for RFC step 3; nothing publishes automation messages
yet.

- **Experiment:** `chat-automations` is in `ExperimentsKnown`,
`ExperimentsUserScoped`, and the ruleguard list in `scripts/rules.go`.
Every route asks `chatd.AutomationsEnabled(ctx, api.ExperimentEvaluator,
callerID)`, which fails closed on a missing subject or evaluator; the
routes stay mounted and return 404 when the experiment is off for the
caller, so a runtime rule for one user applies on the next request
without a restart. There is no `/api/v2` alias.
- **Routes:** list, create (201), get, `PATCH`, delete (204), with
Swagger annotations and audit entries. List returns the automations in
the organization the caller can read (members see their own,
organization admins see all).
- **Service** (`coderd/x/chatd/automations.go`): create validates kind
and target shape, a non-empty prompt, a five-field cron and IANA time
zone, model access, and that an `existing_chat` target is a root chat of
the owner in the same organization. It defaults `when_busy` to `skip`
for schedules and `queue` for webhooks and computes the first schedule
cursor. `PATCH` runs the same per-field validation on each changed
field, rejects `kind` and `target_mode`, bumps `schedule_revision` on
prompt, cron, or time zone changes, and recomputes the cursor on cron or
time zone changes.
- **Owner-only:** only the owner can `PATCH` an automation, even with
organization-admin or site-owner roles (403); admins can list, read, and
delete.
- **Secrets:** a webhook's secret (`coder_automation_` prefix) is
returned once in the create response. Only its SHA-256 is stored, and it
appears in no list, get, audit, or log output.
- **Cap:** new option `--chat-max-automations-per-owner` (default 50)
caps automations per owner across all organizations; creating one more
returns 409. Creates serialize on a per-owner advisory lock.
- **Cron:** `cron.Standard(spec, timeZone)` accepts any standard
five-field expression (the existing `Weekly` rejects day-of-month and
month restrictions). Responses include the next five run times of an
enabled schedule.

PR 3b (stacked on this one) adds disable and re-enable with the queue
cutoff, makes delete run disable first, and adds secret rotation and the
schedule preview route. Until then, the queue promotion guard from
#30096 discards queued rows whose automation was deleted.

Dogfood UAT ran on a real dev instance and passed:
create/list/get/patch/delete via curl, owner-only 403s for an
organization admin and the site owner, the cap (including concurrent
creates and cross-organization counting), a runtime rule turning the
experiment off for one user without a restart, and `/api/v2` 404s.

Part of CODAGT-1209 (PR 3b closes it)
Part of CODAGT-1135
Refs CODAGT-1217

Stack: PR 3 of the Automations for Coder Agents stack, on top of #30096
(which is on top of #30085).

> This PR was written by an AI agent (xum) and reviewed by the agent's
own test audit and UAT. A human has not reviewed it yet.

---
_Generated with [`xum`](https://github.com/coder/xum) • Model:
`anthropic:claude-opus-5-5` • Thinking: `high`_

---------

Signed-off-by: Thomas Kosiewski <tk@coder.com>
2026-10-02 15:54:49 +00:00
Nick Vigilante 2e0f97e5de feat: add type and secret status to the configuration reference (DOCS-896) (#29310)
## Summary

The generated configuration reference told a reader how to set each
deployment option but not what the option accepts. Every entry carried
the environment variable, CLI flag, YAML key, and default, while
serpent's value type stayed in the source. A reader could not tell a
`duration` from an `int`, or discover an enum's accepted values, without
following the flag link into the CLI reference.

- **Type** is now rendered for all 200 options from the serpent value.
Enums expand into their choices (`- Type: `enum`, one of `password`,
`awsiamrds``) rather than printing serpent's escaped `enum[a\|b]`
spelling, which is meant for its own table output.
- **Secrets** are now marked. The nine options carrying the `secret`
annotation that `codersdk.IsSecretDeploymentOption` reads are
deliberately excluded from YAML, so the page now says where to set them
instead of leaving the omission unexplained.

## Correction to the issue

DOCS-896 claimed the page showed a YAML key for secret options. It does
not: all nine secret options have an empty `YAML` field, so the
generator already omitted that bullet. The real gap was that nothing
explained *why* the YAML key is missing. The page header and the
per-option marker now cover it.

## Verification

- `go test ./scripts/configdocgen` passes, with new cases for the
enum-choice parsing, plain and nil value types, and the secret marker
rendering exactly once for the annotated option.
- Regenerated page has 200 `- Type:` lines for 200 rendered options and
9 secret markers.
- `make lint/emdash`, markdownlint, `markdown-table-formatter`, and Vale
are clean on the regenerated page. `golangci-lint` clean on the package.
Full `make pre-commit` passed.

<details>
<summary>Local pre-push note</summary>

The repository pre-push hook runs the full Go suite locally. In this
workspace it segfaults the `make` process outright, and an isolated
`make test` run produces ~1,000 failures unrelated to this change
(embedded NATS servers failing to start within 10s, chatd state-machine
tests), which reproduce independently of this diff. I verified the
change with `go build ./...`, the package tests above, and the full
pre-commit suite instead, and disabled the hook for this worktree only
using its documented `git config coder.pre-push` knob, restoring the
setting immediately after the push. No `--no-verify`. CI is the
authority on the full suite here.

</details>

---

Generated with Coder Agents on behalf of @nickvigilante. Linear:
[DOCS-896](https://linear.app/codercom/issue/DOCS-896/configdocgen-render-option-type-and-secret-status)
2026-10-01 16:50:42 -04:00
Nick Vigilante 4a184ca54a feat: add generated API key scopes reference (DOCS-135) (#29307)
## Summary

Coder validates every requested API key scope against a curated public
catalog in `coderd/rbac`, but the docs never listed it. Readers had to
infer the requestable set from the `codersdk.APIKeyScope` enum, which is
a superset that includes internal scopes a token cannot carry.

This adds `scripts/scopesdocgen`, which renders
`docs/reference/api-key-scopes.md` from `rbac.ExternalScopeNames` and
the policy definitions behind it:

- built-in scopes (`coder:all`, `coder:application_connect`)
- composite `coder:*` scopes, each with the permissions it expands to
- low-level `resource:action` scopes, with the action descriptions the
policy declares
- the deprecated `all` / `application_connect` aliases and their
canonical spellings

The generator is wired into `make gen`, so CI fails when the scope
catalog and the page drift apart.

## Also in this change

- `coderd/apikey.go`: the unrequestable-scope `400` detail pointed at
`admin/users/sessions-tokens#api-key-scopes`, which only ever held a
partial list. It now points at the generated reference.
- `docs/admin/users/sessions-tokens.md`: the partial enumeration and the
"read the schema, but not all of it applies" caveat are replaced with a
link to the reference. The how-to content stays.

## Verification

- `go test ./scripts/scopesdocgen` covers the sentence formatting, the
scope partitioning, and a guard that every name in
`rbac.ExternalScopeNames()` appears on the rendered page.
- `go test ./coderd -run
'TestTokenScoped|TestCreateTokenScopes|TestExternalScopesAreStorable|TestTokenLegacySingularScopeCompat'`
passes, confirming the page's claims about the default scope and the
rejection of internal scopes.
- `make lint/emdash`, `make lint/docs-html`, `make lint/check-scopes`,
`pnpm run check-docs`, and Vale on the changed pages are clean. Full
`make pre-commit` passed.

<details>
<summary>Context and scope decisions</summary>

Part of the DOCS Automated reference documentation project, which tracks
generating reference pages from product source of truth instead of
maintaining them by hand. DOCS-135 is the first item: it was filed as
"wire up the existing generator," but `scripts/apikeyscopesgen` emits Go
(`codersdk/apikey_scopes_gen.go`), not documentation, so there was no
generator output to surface and this adds one.

Decisions made while implementing:

- **Placement.** `docs/reference/api-key-scopes.md`, as a direct child
of Reference rather than inside `docs/reference/api/`, because that
directory is owned by the swagger generator.
- **Page shape.** Follows the `scripts/docgenenv` convention for
generated reference pages: the generated marker and page metadata in
front matter, the generated-content banner in the body, and sections
starting at level two because the front-matter title renders as the page
heading.
- **Public scopes only.** The page lists what `rbac.IsExternalScope`
accepts. Internal `resource:action` pairs are deliberately excluded,
which is the distinction the old schema link failed to make.
- **Acronym handling.** Policy action descriptions are lowercase source
strings (`"ssh into a given workspace"`). The generator restores the
casing of a small set of leading acronyms rather than editing the RBAC
package, so the change stays inside the docs pipeline.

Sibling backlog items turned out to be already delivered and are being
rescoped separately: `metricsdocgen` and `auditdocgen` already write
into their docs pages, and `scripts/configdocgen` already generates the
configuration reference.

</details>

---

Generated with Coder Agents on behalf of @nickvigilante. Linear:
[DOCS-135](https://linear.app/codercom/issue/DOCS-135/wave-1-wire-up-api-key-scopes-reference-to-docs)


## Product UI impact, please read

This PR corrects three RBAC action description strings in
`coderd/rbac/policy/policy.go`:

| Action | Before | After |
|---|---|---|
| `workspace:start` | allows starting a workspace | start a workspace |
| `workspace:stop` | allows stopping a workspace | stop a workspace |
| `api_key:update` | update an api key, eg expires | update an api key,
for example its expiry |

These were internal-facing until this PR, which is what made them worth
fixing: the new reference page renders them verbatim to operators, and
"eg expires" is not decodable by a reader without codebase context.

**They are not docs-only.** `make gen` propagates them to
`coderd/rbac/object_gen.go` and
`site/src/api/rbacresourcesGenerated.ts`, and
`CreateEditRolePageView.tsx` renders the description next to each action
in the custom roles editor. So this changes admin UI copy, and
`CreateEditRolePageView.stories.tsx` means Pixel may now report a
genuine snapshot diff on that component rather than the pre-existing
failure noted below.

The new wording matches the convention the rest of the table already
follows (`ssh into a given workspace`, `create an AI Gateway key`), so
this makes the UI more internally consistent, not less. Flagging it
explicitly because a docs PR changing product copy deserves an owner's
eyes rather than a silent merge. Happy to split it out if RBAC or
frontend would rather own it, with the caveat that the reference page
would then publish `eg expires` in the meantime.

<details>
<summary>CI note: Storybook / Pixel failure</summary>

The `Storybook` check fails at its `Snapshot` step, and `Pixel / Review`
reports `build failed` with no snapshots processed.

**Updated:** this note originally said the branch touched no files under
`site/`. That is no longer true. Correcting the RBAC policy description
strings (CRF-24) required `make gen`, which regenerated
`site/src/api/rbacresourcesGenerated.ts` alongside
`coderd/rbac/object_gen.go`. That file is a generated string table with
no rendered component behind it, so it still cannot account for a
Storybook snapshot or Pixel build failure, but the original
justification no longer holds and is replaced by this one. The failure
predates that change: it was present when the branch genuinely had no
`site/` diff. Every other check passes.

</details>
2026-10-01 16:09:01 -04:00
Nick Vigilante 149366a459 feat: validate documented experiment keys (DOCS-139) (#29316)
## Summary

The docs name specific experiment keys wherever an individual
experimental feature is documented, and nothing verifies those keys
still exist in the code.
When an experiment is renamed or graduated, the enablement example keeps
telling operators to set a key the server no longer accepts.

This adds `scripts/checkexperimentkeys`, which fails when a key named in
`docs/**` is absent from `codersdk.ExperimentsKnown`, and wires it into
`make lint` as `lint/check-experiment-keys`.

**The data flows one way.**
The check reads keys the docs already name and validates them against
code.
It does not generate a page, add a manifest route, or write any key into
`docs/`.
That is deliberate: #29113 retired the generated feature-stage lists
because the team decided not to advertise which features are currently
experimental or beta.
A guard against stale keys is compatible with that decision; an index is
not.

The five key references the check currently guards are the per-feature
ones that decision left in place: `chat-advisor`,
`chat-virtual-desktop`, `auto-fill-parameters`, `agent-lifecycle-hooks`,
and `mcp-server-http`.

### Recognition rules

Only explicit enablement forms are matched: `--experiments=<key-list>`
and `CODER_EXPERIMENTS=<key-list>`, including every key in a
comma-separated list.
Values may be single- or double-quoted.
Explicit shell line continuations are joined before matching, preserving
the first physical line for a finding.
Bare identifiers in prose are ignored on purpose.
A check that guesses at prose produces false positives, and a check with
false positives gets deleted by the first person it blocks.
`*`, `feature1`, `feature2`, and angle-bracketed values are recognized
placeholders.
The recognized forms are documented in the script's README.

Known gap: the space-separated `--experiments <key>` form is not
matched.
No prose uses it today, so this is latent rather than broken, and
closing it is a small follow-up if the form reappears.

### Also fixed

`ExperimentMCPToolSearch` had no case in `DisplayName()`, so it rendered
as a title-cased guess in the product UI as well as anywhere else
display names surface.
`TestExperimentDisplayNames` now pins the display name of every known
experiment against an expected map, with a length assertion, so a new
experiment cannot silently fall back to the title-cased key.

## Verification

- `go test ./scripts/checkexperimentkeys ./codersdk` passes.
Coverage includes known and unknown keys, quoted values, shell
continuations, placeholders, multi-key lists, bare-prose exclusion, and
failure-output formatting.
- `golangci-lint run ./scripts/checkexperimentkeys/... ./codersdk/...`
reports 0 issues.
- `make lint/check-experiment-keys` is clean against current `docs/`.
- Negative control: a fixture containing
`CODER_EXPERIMENTS=not-a-real-key` exits 1 with the expected finding.
- The full `pre-commit` hook passes, including
`lint/check-experiment-keys` inside aggregate `make lint`.

<details>
<summary>History: the deferred lint wiring, now resolved</summary>

This PR previously landed the tool and a standalone `make` target while
leaving the one-line addition to aggregate `make lint` for a follow-up,
and stayed draft, because three documented `oauth2` keys were stale and
owned by another PR:

- `docs/admin/integrations/oauth2-provider.md:34`
- `docs/ai-coder/mcp-server.md:108`
- `docs/ai-coder/mcp-server.md:114`

#29268 merged on 2026-09-15 and removed all three in favor of
`CODER_OAUTH2_PROVIDER_ENABLE`.
The branch is rebased onto that, the check is green against `main`, the
wiring is folded into this PR, and the merge-order constraint recorded
as CRF-5 no longer applies.

The rebase also collided with `DisplayName()`, which gained new cases on
`main`.
Resolved in favor of `main`'s switch rather than this branch's earlier
map refactor, keeping the production diff to the one missing case and
moving the exhaustiveness expectations into the test.

</details>

<details>
<summary>Why this replaced a generated reference page</summary>

DOCS-139 originally asked for a generated
`docs/reference/experiments.md` listing every key in `ExperimentsKnown`.
That was built and opened as #29311, then closed without merging.

Commit b23d73d5 (#29113, six days earlier) had removed both generated
feature-stage indexes from `docs/install/releases/feature-stages.md`,
recording that the decision, after discussion with David, was not to
advertise either list, and describing them as "drift surfaces that
promise a completeness the docs cannot maintain."
It explicitly rejected the superseded #28949, which generated the list
from `ExperimentsKnown` with descriptions parsed from Go constant
comments, for making docs track "an internal, unstable surface."

#29311 was that rejected design rebuilt.
The ticket predated the decision; the page should not have been built on
top of it.
This PR is the part of the original goal that survives: the drift
protection, without the publication.

</details>

<details>
<summary>Local pre-push note</summary>

The pre-push hook's test stage cannot run in this workspace: GNU Make
4.4.1 segfaults on the repo's `.SHELLFLAGS = $@ -ceu` configuration for
a target named `test`, tracked in DOCS-903 with a minimal reproduction.
Verified instead with the package tests, `golangci-lint`, and the full
pre-commit hook.
No `--no-verify`.

</details>

---

Generated with Coder Agents on behalf of @nickvigilante. Linear:
[DOCS-139](https://linear.app/codercom/issue/DOCS-139/validate-documented-experiment-keys-against-experimentsknown)
2026-10-01 20:08:40 +00:00
Nick Vigilante 72e31ffb4c fix: prevent Prometheus metrics reference drift (#29314)
## Summary

The Prometheus metrics reference merged scanner output with a
hand-maintained file, and changes to the hand-maintained half could not
fail CI. 66 of the 76 static families had no scanner coverage: four
files sat in a skip list pending `WrapRegistererWithPrefix` support, and
the `aibridge/` tree was outside the scan roots entirely. Rename or add
a metric in any of them and `make gen` regenerated the page unchanged,
so `check_unstaged.sh` passed while the published table quietly went
wrong.

The scanner now resolves the prefixes a registerer applies:

- Indexes `WrapRegistererWithPrefix` and `NewMetricAliasRegisterer` call
sites, resolving the prefix from a string literal or an imported
constant.
- Maps that prefix to the package whose constructor receives the wrapped
registerer, covering the two shapes the codebase uses: a constructor
that **forwards** its registerer to another package
(`aibridge.NewMetrics` to `aibridge/metrics.NewMetrics`), and a
collector **registered directly** against the wrapped registerer
(`gatewayRegisterer.MustRegister(keypool.NewStateCollector(...))`).
- Indexes only the canonical prefix of an alias registerer. The legacy
`coder_aibridged_` and `coder_aibridgeproxyd_` aliases are deprecated
per AIGOV-447 and should not be taught in the reference.

With prefixes resolved: skip list deleted, `aibridge` added to the scan
roots, and all 29 `coder_ai_gateway_*` families removed from the static
file, which now holds only genuine externals (`go_*`, `process_*`, and
metrics built dynamically). `enterprise/scaletest/agentfake` is excluded
deliberately: its fake-agent metrics are not part of a deployment's
surface, so documenting them would send operators looking for series
they will never see.

**Also fixes two extraction and merge gaps.** Label lists written as
`append(baseLabels, "status", "route")` now resolve, and any
unresolvable element returns no labels rather than a partial list.
Static runtime-label overrides now replace generated labels only:
generated name, type, and HELP remain authoritative.

## Verification

The prefix-scanning refactor itself preserves the existing AI Gateway
rows. However, the page is **not byte-identical** to `main` because this
round fixes two already-published stale HELP strings. Its SHA-256
changes from
`7ced1e7e9f2d4caf5e3b5e484ef4af14f955b8b4cd1a14fe81360bbdd49bb150` to
`a68774275a3b21d5bae65024cc8ee65b52c51e52786c9aea45e3981e4e7f1643`.

| Metric | Before | After |
| --- | --- | --- |
| `coderd_agentstats_connection_median_latency_seconds` | The median
agent connection latency | The median agent connection latency in
seconds |
| `coderd_agentstats_startup_script_seconds` | The number of seconds the
startup script took to execute. | Amount of time taken to run the
startup script in seconds. |

- `go test ./scripts/metricsdocgen/...` covers prefix resolution from a
literal and imported constant, alias-registerer argument selection,
collector registration, forwarder propagation, scan-to-emit prefix
application, all-or-nothing label extraction, static-label merging, and
the exclusion predicate.
- `golangci-lint run ./scripts/metricsdocgen/...` and full `make
pre-commit` pass.

## What this does not fix

Ten metrics are still skipped for having no `HELP` text, with a warning
to stderr rather than an error. Making that fatal means writing accurate
descriptions for ten metrics in the agent and SSH server packages, which
is a behavioral claim I would rather not guess at. Follow-up filed as
DOCS-906.

The scanner still uses package-level syntactic prefix inference. Broader
validation of prefix scan roots, mixed registration within a package,
and parameter-passed registerers belongs in DOCS-908 rather than this
focused drift fix.

<details>
<summary>Local pre-push note</summary>

The pre-push hook cannot run in this workspace: GNU Make 4.4.1 segfaults
on the repo's `.SHELLFLAGS = $@ -ceu` configuration for a target named
`test`, tracked in DOCS-903 with a minimal reproduction. I verified with
the package tests, `golangci-lint`, and full pre-commit, and disabled
the hook for this worktree only through its documented `git config
coder.pre-push` knob, restoring it right after the push. No
`--no-verify`.

</details>

---

Generated with Coder Agents on behalf of @nickvigilante. Linear:
[DOCS-897](https://linear.app/codercom/issue/DOCS-897/metricsdocgen-static-metrics-file-can-go-stale-without-failing-ci)
2026-10-01 16:06:32 -04:00
Susana Ferreira 35abc22fa9 fix(scripts/aibridgepricesgen): drop claude-sonnet-5-5 bare ID injection (#30229)
The scheduled `aigateway-prices-refresh` workflow fails
([run](https://github.com/coder/coder/actions/runs/36840685231)) with
`anthropic.claude-sonnet-5-5 now present upstream; drop the injection`.

#30159 added an `overrides.jq` injection for the Bedrock ID
`anthropic.claude-sonnet-5-5`, copied from
`global.anthropic.claude-sonnet-5-5`, because models.dev only listed the
global profile at the time. A few hours later, models.dev added the bare
ID itself
([sst/models.dev#8417](https://github.com/sst/models.dev/pull/8417))
with the same pricing and limits, so the guard now fails the refresh.

This removes the injection. `prices.json` is unchanged: the existing
`anthropic.claude-sonnet-5-5` row now comes from upstream with the same
values. Other upstream data drift is left to the workflow's own refresh
PR.

After merge, `aigateway-prices-refresh` will be rerun manually on `main`
to open the refresh PR.

> [!NOTE]
> Created by Coder Agents on behalf of @ssncferreira.
2026-10-01 11:05:28 +01:00
Nick Vigilante f1ed84ffab docs: author the sidebar in per-section YAML compiled to manifest.json (#30165)
## Summary

The docs sidebar is now written as YAML files under `docs/manifest/`,
one route and its direct children per file, and a new tool,
`scripts/docsmanifestgen`, compiles them into `docs/manifest.json`.
Moving two top-level sections is now a 2-line change in
`docs/manifest/index.yml` instead of a diff of hundreds of lines in a
2,600-line JSON file.
The compiled `docs/manifest.json` is byte-identical to `main`, so
coder.com, docs-preview, deploy-docs, and offlinedocs need no changes.

Refs
[DOCS-1131](https://linear.app/codercom/issue/DOCS-1131/author-the-docs-sidebar-in-per-section-yaml-compiled-to)

## Changes

- **Sources:** `docs/manifest/index.yml` holds `versions` and the
section files in sidebar order.
Each section file (`about.yml`, `admin.yml`, and so on) holds one
top-level route.
- **One level per file:** a file holds one route and its direct
children.
A child with children of its own, or with `children_from`, lives in its
own file in a directory named after the parent file, and the parent
lists it as `- include: admin/users.yml`.
For example, `get-started.yml` includes
`get-started/customize-your-template.yml`, which lists that page's
children.
  The 8 section files split into 53 source files.
- **`scripts/docsmanifestgen`:** `build` compiles the sources into
`docs/manifest.json`, `fmt` formats them, and `check` validates them.
- Keys must be known (a typo like `titel:` fails with a line number) and
appear in a fixed order in both the YAML and the JSON.
- Every `path` must exist, and every file under `docs/manifest/` must be
referenced from `index.yml`, an `include`, or a `children_from`.
- A grandchild in the same file fails with a message to move the child
into its own file, and an `include` entry can't set other keys or name a
file that is already used.
- A comment on the same line as a `title` or `description` is rejected,
because an unquoted ` #` would silently cut off the value.
- **Generated fragments:** `clidocgen` and `apidocgen` no longer rewrite
the manifest.
They write the Command Line and REST API children to
`docs/manifest/generated/*.json`, which the sources pull in with
`children_from`.
  The tool accepts fragments as YAML or JSON.
- The hand-written REST API descriptions and states moved into the YAML,
attached to generated routes by title.
A YAML entry that matches no generated route, or disagrees with one,
fails the build instead of being dropped silently.
- **Shared code:** loading, validation, and fragment merging live in
`scripts/docgenenv/sources.go`, so the generators and the tool share one
loader.
`docgenenv.Route` gains YAML tags and the source-only `ChildrenFrom` and
`Include` fields (`json:"-"`).
`LoadManifest` is removed because nothing reads the compiled JSON
anymore.
- **Makefile:**
- `docs/manifest.json` is built from the sources and fragments, which
replaces the `_gen/manifest-staging.json` handoff.
- Each fragment has its own rule, and the fragments are added to the
`.PRECIOUS`, `GEN_FILES`, and `gen/mark-fresh` lists.
  - `DOCS_MANIFEST_SOURCES` finds the source files recursively.
- New targets: `fmt/docs-manifest` (part of `fmt`), `lint/docs-manifest`
(part of `lint`), and `gen/docs-manifest`, the quick path after a
sidebar-only edit that skips regenerating the CLI and API fragments.
- **`.gitattributes`:** marks `docs/manifest.json` and
`docs/manifest/generated/*` as generated so GitHub collapses them in
diffs.
- **Guidance:** the content guidelines, the style guide,
`DOCS_STYLE_GUIDE.md`, the `write-docs` and `doc-check` skills,
`docs/README.md`, and `docs/tutorials/example-guide.md` now point to the
YAML sources instead of `docs/manifest.json`.

## Testing

- Split today's `docs/manifest.json` into the sources, rebuilt it, and
confirmed the output is byte-identical to `main`.
After splitting the sources to one level per file, the rebuilt
`docs/manifest.json` and a fresh `clidocgen` CLI fragment were still
byte-identical, and `LoadManifestSources` still finds the REST API and
Command Line routes through their includes.
After rebasing onto `main`, both commits rebuild `main`'s
`docs/manifest.json` byte for byte.
- Hand-edited `docs/manifest.json` and a YAML source, then rebuilt: the
YAML change persisted and the JSON edit was overwritten.
CI's `make -B gen` plus `check_unstaged.sh` fails on a committed hand
edit to the JSON.
- Deleted both fragments and ran `make docs/manifest.json` in a fresh
worktree, which regenerated them through swag and `clidocgen`.
`docs/manifest.json`, the fragments, and every generated CLI and API
page had no diff.
- Swapped User Guides and Administration in `index.yml`: 2 changed lines
in the source, 166 generated lines in `docs/manifest.json`.
  Swapping back restored the file exactly.
- `go test` for `scripts/docgenenv`, `scripts/docsmanifestgen`,
`scripts/clidocgen`, and `scripts/apidocgen/postprocess`.
The new tests cover merging, includes, every validation error,
idempotent formatting that keeps comments, and a check that the key
order matches the struct field order.
- `golangci-lint` on the changed packages, and the full `make
pre-commit` hook (gen, fmt, lint including `lint/docs-manifest`, and the
slim build).

## Notes

- Open PRs that edit `docs/manifest.json` will need to redo their
sidebar changes in the YAML sources after this merges.
- #30161 (merged) stopped escaping `&`, `<`, and `>` in
`docs/manifest.json`.
This PR carries that over: all manifest JSON escaping is in
`docgenenv.MarshalJSON`, which sets `enc.SetEscapeHTML(false)`, and
`Manifest.Marshal` from #30161 is gone because the generators no longer
write the manifest.
- #30129 (merged) flattened `admin/external-auth/index.md` and
`ai-coder/agents/tools/index.md` to single files, and the YAML sources
use the new paths.
- The CLI index and REST API page front matter still update only on a
full `make gen` when their sidebar metadata changes, as it did before.
That gap and the overall cost of `make gen` are tracked in
[DOCS-1141](https://linear.app/codercom/issue/DOCS-1141/reduce-the-compute-cost-of-make-gen).

<details>
<summary>Decision log</summary>

### Goal

Make the sidebar easy to reorder and hard to break, with strict key
order and YAML sources, while keeping `docs/manifest.json` as the file
downstream systems read.

### Jsonnet vs. a custom tool

Jsonnet was evaluated first and rejected, based on tests against
go-jsonnet v0.22.0:

- Jsonnet always sorts object keys alphabetically, and there's no option
to keep them in order (`preserve_order` isn't a parameter of
`manifestJsonEx` or `objectFieldsEx`).
Passing today's manifest through Jsonnet and then biome changed 2,114 of
2,478 lines.
- Unknown keys pass through silently, so a `titel:` typo ends up in the
output.
- The sources would be `.jsonnet` files, not YAML.
Reading YAML takes `std.parseYaml(importstr ...)`, and the keys still
come out sorted.
- Restoring key order would need a Go step after Jsonnet anyway, and
that step is most of this tool.

Decoding YAML into Go structs with yaml.v3 (already a dependency) gives
key order from the struct, rejects unknown keys, keeps comments, and
reproduced the current manifest byte for byte.

### Design choices

- **Ordered by `index.yml`:** reordering sections touches only the
index.
- **One route and its direct children per file:** every file stays short
and flat, so a subsection can be found, reviewed, and moved as one file.
Files nest by parent file name rather than by docs path, so the
directory tree matches the sidebar tree.
Include paths are relative to `docs/manifest/`, like `children_from`, so
every reference to a file is spelled the same way.
A child with `children_from` also gets its own file, so the rule has no
exceptions.
- **Generated subtrees in their own fragments:** the generators can't
safely rewrite hand-edited YAML, so they own separate files.
Hand-written metadata for generated routes lives in the YAML, attached
by title, so descriptions never live in a generated file.
- **Key order enforced in both sources and output:** `fmt` rewrites the
sources, `check` fails on unformatted sources, and the output order
comes from the `docgenenv.Route` field order.
- **`&` left unescaped:** the output matches `main` after #30161, which
stopped escaping `&` downstream.
- **`gen/docs-manifest` separate from `fmt`:** keeping manifest assembly
out of `fmt` means `make -j gen fmt` never has two jobs writing
`docs/manifest.json` at once.
- **YAML pitfalls:** unquoted `no` and `123` stay strings because the
fields are typed as strings, an unquoted `: ` is a parse error, and the
silent ` #` truncation is rejected by the comment check.

### Out of scope

- Replacing `&` with "and" in titles (a style question).
- Reducing the cost of `make gen` (DOCS-1141).

</details>

---

_This PR was generated by Coder Agents on behalf of @nickvigilante._
2026-09-30 21:27:14 +00:00
Nick Vigilante 0ce91ffc18 fix: stop escaping ampersands in docs manifest titles (#30161)
Closes DOCS-1139.

Go's `json.MarshalIndent` escapes `&`, `<` and `>` for HTML safety, so
both docs generators wrote titles like "Licensing & usage" into
`docs/manifest.json` as `Licensing & usage`. The two forms decode
identically (the offline docs site reads the file with `JSON.parse`),
but the escaped form reads like a bug in review, and a hand-typed `&`
flipped back on the next `make gen`.

Add `Manifest.Marshal` in `scripts/docgenenv`, which disables HTML
escaping, and use it from `clidocgen` and `apidocgen/postprocess`. The
two titles in `docs/manifest.json` that still carried the escape are
updated to match. A unit test covers the unescaped output and the round
trip through `LoadManifest`.

Testing: `go test ./scripts/docgenenv ./scripts/clidocgen
./scripts/apidocgen/...` passes. I ran `clidocgen -manifest-only`
against a copy of the manifest and confirmed the output has no escapes
and parses to the same content as the committed file. I did not run the
full `make docs/manifest.json` (it needs the site dependencies and
Biome), so the committed file keeps its existing Biome formatting.

cc @ianjevans for awareness.

---
🤖 Built with AI assistance.
2026-09-30 14:35:36 -04:00
J. Scott Miller fc0af7fbd8 feat: open a chat_turn span per prompt (#29771)
**TL;DR:** Wraps each prompt in a `chat_turn` root span with every
lifecycle stage parented under it. The span records one outcome per turn
(`completed`, `interrupted`, `error`, `abandoned`), and the new
`coderd_chatd_turn_outcomes_total` counter counts those outcomes. Turns
stay correct when tasks are canceled, replaced, or retried.

Each prompt runs inside a `chat_turn` root span owned by the runner
(`runnerTurnSpan`). The generation step, prepare, MCP connect, stream,
TTFT, provider attempt, commit, thinking, tool call, compaction, and
retry backoff stages are parented under it, activating the stages from
#29770. `chat_turn` spans are standalone roots: the turn runs on a
worker, often on another replica, and no trace context is persisted.

**Lifecycle**

- **Start.** Each generation loop iteration calls `Ensure`, which opens
the turn if none is open and returns a `turnToken`. The span is
backdated to the trigger: the latest of the last user prompt, a dynamic
tool's result message after it, and a pending compaction request. An
`acquisition` stage covers trigger to pickup.
- **Stale trigger / takeover.** A trigger at or before the previous
anchor opens the turn at now with no `acquisition` and counts
`stale_anchor`. A runner that took the chat over from an owner with a
stale heartbeat (`TakenOver`, from acquisition) opens its first turn at
now with no `acquisition`, so the dead owner's run is not reported as
pickup delay.
- **Newer prompt.** `Ensure` on the current turn with a newer trigger
closes it and opens a new one. The closed turn is `abandoned` unless the
task that ran it records an outcome before releasing it.
- **Finish.** After `FinishTurn` or `EnterRequiresAction` commits, the
step calls `Complete`; `Settle` closes the span after the step's own
stage ends, so the finishing step is inside the turn. A chat waiting in
`requires_action` is outside any turn; the generation resumed by its
tool results opens a new one.
- **Emission.** Closing a turn fixes its end time; its span is emitted
once every task that joined it (`Ensure` takes the task ID) has called
`Release`, which `runTask` does after the task returns. A finishing task
whose `FinishTurn` promoted a queued prompt therefore still counts its
turn as `completed` even when the promoted task's `Ensure` runs before
its `Complete`. `End` at runner exit emits whatever is left.
- **Failure.** A step that commits `FinishError`, or fails closed,
invalidates the turn as `error` after the commit. Errors the task runner
retries, including task timeouts, keep the same turn open; an expected,
non-retryable exit records no outcome and leaves the turn open until the
interrupt task, a newer prompt, or runner exit closes it. A canceled
generation task records no outcome and cannot obtain a turn token, so a
replaced task never touches its successor's turn.
- **Interrupt.** The interrupt task marks the open turn `interrupted`
before `FinishInterruption` commits and closes it after. Only a stop (by
the user or a parent agent) is `interrupted`; it ends with unset span
status, since error status is reserved for `error` turns. runner
shutdown, loss of ownership, and message edits close the turn as
`abandoned` with unset status.
- **Promotion.** Every promoting transition (`FinishTurn`,
`FinishInterruption`, a send to an errored chat, `PromoteQueued`)
returns the promoted message's queued time in its result, and one
`recordQueueWait` helper records a standalone `queue_wait` to the
promotion. The promoted turn opens at the promotion time.

`chat_turn` is observed on `coderd_chatd_stage_duration_seconds` only
for completed turns, so the histogram is completed-turn latency; every
turn still emits its span.

`coderd_chatd_turn_outcomes_total{outcome, chat_kind}` counts every
closed turn exactly once (`completed`, `interrupted`, `error`,
`abandoned`), matching the span's `turn_outcome`, under the
`chat-stage-metrics` experiment. Turn time accounting is in draft
#29772.

**Stages and identity**

- `tool_call` is a live stage around each local tool call in `chatloop`;
the tool runs on its context, and it ends errored only when the tool
fails to execute. Calls that never run (inactive, unknown, or rejected
by policy) record no stage.
- `mcp_connect` carries connected and failed server counts and ends
errored only when servers were configured and none connected.
- The compaction stage carries the model that runs the summary.
- `withStageIdentity` puts scope and chat kind on the turn context.
Detached background work runs in `background` scope as its own trace
roots.
- The `advisor` tool's nested model call appears only as its enclosing
`tool_call`.

ARCHITECTURE.md gains a "Lifecycle tracing" section.

<details>
<summary>Review decisions</summary>

- Turn identity stays timestamp-based with targeted fixes (clamped
acquisition, rotation on a newer trigger, invalidate after commit,
fail-closed invalidation, settle closing invalidated turns) plus a
takeover flag; keying turns on message IDs would not detect takeover and
cannot express interrupted/abandoned outcomes.
- Outcome classification: the interrupt task owns `interrupted`, and
cancellation or a fence mismatch alone never records an outcome, so
shutdown, ownership loss, and edits are `abandoned`. Only `error` turns
set error span status. Remaining ordering races between a step and an
interrupt are documented as known limitations in ARCHITECTURE.md.
- Turns are emitted after the tasks that ran them return, not when they
close, so an outcome a replaced task records while unwinding reaches its
own turn; the successor task is not delayed. Chosen over an
after-commit, pre-publish hook on `ChatMachine.Update`, which would fix
only the `FinishTurn` ordering.
- `requires_action` ends the turn rather than pausing it, so `chat_turn`
measures chatd time; a prompt with client tools counts as several turns.
- Promoted turns anchor at promotion time and `queue_wait` is always
standalone, so `chat_turn` never includes time spent queued behind
another turn.
- The former `capacity_wait` stage (#29805) was dropped: it was a
per-worker lower bound, excluded from accounting, and needed its own SQL
and bookkeeping. #29805's remaining helpers are folded into this PR.

</details>

Supersedes #28972, #28973, and #29805.

🤖 This PR was generated by Coder Agents on behalf of @jscottmiller.
2026-09-30 11:53:58 -05:00
J. Scott Miller e988ddb11d feat: trace chatd provider stages and count only output parts as first token (#29770)
**TL;DR:** Traces the model provider call: a `stream` stage, a nested
`time_to_first_token` stage, and one `provider_attempt` per HTTP round
trip, labelled with provider type and model. Behavior change:
`coderd_chatd_ttft_seconds` now counts only a real output part as the
first token, so error, warning, and finish parts no longer count.

Instruments the provider call.

**Stream and time to first token.** `GenerateAssistant` records a
`stream` stage from stream open to close and a nested
`time_to_first_token` stage. The TTFT window closes once, on the first
streamed output part: any part other than warnings, finish, or error,
including start markers such as `text_start`. A closed window observes
the stage and the pre-existing `coderd_chatd_ttft_seconds` with the same
elapsed time. A failed open, an error part, or a release without an
output part ends the TTFT span with the error (a silence timeout is
classified as such) and observes nothing. The `stream` span ends once,
with the error `GenerateAssistant` returns, so content-filter and
no-output attempts end errored. Failed `stream` windows are observed
like successful ones. The attempt is released before the stream span
ends, so the TTFT window always falls inside its stream. The compaction
summary stream passes a nil tracer, so it records neither.

Without #29771, production emits only `provider_attempt`; #29771 passes
the tracer to `GenerateAssistant`.

**Provider attempts.** The provider HTTP transport is wrapped in
`stageSpanRoundTripper`, which emits one `provider_attempt` stage per
round trip, so retries each get a span. The span closes on response
headers; a status >= 400 marks it errored without changing what the
client sees.

`resolveModelCall` resolves the `StageModel` (wire provider, configured
`provider_type`, model, effective reasoning effort) and threads it to
the transport, so span attributes and labels match the call.
`provider_type` is read from `ai_providers.type`, so types outside the
fantasy set (copilot) keep their configured name. `chatd.New` builds the
tracer with the server clock.

**Behavior change to `coderd_chatd_ttft_seconds`:** only an output part
counts as a first token. Error parts no longer close the window as a
success, and warnings and finish parts are skipped; the help text says
so.

`StageSpan.End` returns the span's elapsed window, which `ttft_seconds`
observes. Callers pass `Stages` and `StageModel` through
`GenerateAssistantOptions`; a nil tracer discards (TTFT still records
`ttft_seconds`). `Server` gains the `stages` tracer and a
`TracerProvider` option, wired from `coderd`.

<details>
<summary>Review decisions</summary>

- Start markers close the TTFT window: they are the first sign the model
is producing output, and waiting for a delta would fold hidden reasoning
time into TTFT.
- Failed `stream` and `provider_attempt` windows are observed: that time
was spent. TTFT is the exception because a first token that never
arrived has no duration.
- `http_host` was dropped from `provider_attempt`; every request goes
through the local AI Gateway transport, so it was constant.
- The advisor tool's nested model call is intentionally not instrumented
as `stream`/TTFT.

</details>

Supersedes #28970 and #28971.

🤖 This PR was generated by Coder Agents on behalf of @jscottmiller.
2026-09-30 11:53:58 -05:00
J. Scott Miller 7a9442014e feat: add chatd lifecycle stage tracer (#29769)
**TL;DR:** Adds a stage tracer that turns each chat lifecycle stage into
both an OpenTelemetry span and a Prometheus histogram observation, so
traces and metrics always agree. The metrics are only registered with
the `chat-stage-metrics` experiment. Nothing calls the tracer until the
next PRs.

Adds `chatloop.StageTracer`, which emits each chat lifecycle stage as an
OpenTelemetry span and a `coderd_chatd_stage_duration_seconds{stage,
scope, chat_kind}` observation from one `End` call, so trace and metric
durations cannot disagree. The provider-bound stages (`stream`,
`provider_attempt`) are also observed on
`coderd_chatd_model_stage_duration_seconds{stage, provider_type,
chat_kind, model}`, the only family carrying `provider_type` and
`model`, and only for turn-scoped stages. Time to first token per model
stays on the existing `coderd_chatd_ttft_seconds`. Nothing calls the
tracer yet; the next PRs instrument the provider call and the runner.

- The families are registered only when the `chat-stage-metrics`
experiment is enabled (safe and known, so
`--experiments=chat-stage-metrics` or `--experiments=*` turns it on).
Spans are emitted regardless. `NewMetrics` leaves the families off;
`chatd.New` opts in from the experiment, and a test asserts the
experiment decides whether they reach the server's registry.
- `Stage`, `Scope`, `ChatKind`, and `StageAnomaly` are typed string
constants. `scope` separates latency attributable to a prompt (`turn`)
from detached `background` work; `chat_kind` is `root` or `subagent`,
empty without a known chat.
- `StageModel` carries the wire `provider`, `provider_type`, `model`,
and `reasoning_effort`. All are span attributes; only `provider_type`
and `model` are metric labels, on the model family. No family carries an
organization label.
- Observed stages: `chat_turn`, `queue_wait`, `acquisition`,
`mcp_connect`, `stream`, `time_to_first_token`, `provider_attempt`,
`tool_call`, `commit`, `retry_backoff`. `generation_step`, `prepare`,
`thinking`, and `compaction` are span-only. A test keeps the stage sets
and help text consistent.
- API: `Start` / `StartRootAt` for live spans (span timestamps come from
the tracer clock), `Record` for stages reconstructed from timestamps
(the scope comes from the context). `End` is idempotent;
`EndWithoutObservation` ends the span without a sample. A nil tracer,
provider, or metrics discards.
- `coderd_chatd_stage_anomalies_total{reason}` counts observations
dropped, adjusted, or inconsistent (`negative_elapsed`,
`inverted_window`, `missing_timestamp`, `future_start`, `stale_anchor`).
Span-only stages never feed it.
- Buckets are 12 round edges (`0.1 0.25 0.5 1 2.5 5 10 30 60 300 1800
3600`).
- Moves the body of the `for` loop in `StartGeneration` into
`runGenerationStep` so the step has a function boundary for a deferred
stage end. Control flow and error wrapping are unchanged.

<details>
<summary>Series cost (approximate, whole stack)</summary>

A histogram label set is 12 edges + `+Inf` + `_sum` + `_count`: 15
series. Per replica, each `chat_kind` costs about 150 stage series and 4
turn outcome counters; each `(chat_kind, model)` costs 30 (2 stages x
15). Both chat kinds on `M` models: about `308 + 60*M` per replica, plus
background-scope stage series.

Dogfood `main`, 3 replicas: ~3.4k `coderd_*` series per replica (~10.1k
total).

| Models used per replica | Added series (3 replicas) | vs. baseline |
|---:|---:|---:|
| 1 | ~1.1k | +11% |
| 5 | ~1.8k | +18% |
| 10 | ~2.7k | +27% |
| 35 | ~7.2k | +72% |

Most deployments run four or fewer models, so a typical replica adds
roughly 370 to 550 series. The experimental turn accounting in #29772
would add 88 more per replica (11 categories x 4 outcomes x 2 chat
kinds). Series appear lazily and do not grow with organizations.

</details>

<details>
<summary>Review decisions</summary>

- Experiment gate only; no deployment flag or metric levels. If the
experiment graduates, the families become always on like the other chatd
families; a flag is added then only if dogfood shows a cardinality
problem.
- Organization is on neither metrics nor spans, and `effort` is
span-only; `provider_type` (configured type), not the wire `provider`,
is the model-family label.
- `time_to_first_token` is not on the model family; per-model TTFT is
`ttft_seconds`, which observes the same measured window.
- The model family records only turn-scoped stages.
- Nil guards are kept so a nil or zero tracer is always safe to call.
- HTTP attribute keys stay snake_case like the rest of the chatd span
attributes.

</details>

Supersedes #28968 and #28969.

🤖 This PR was generated by Coder Agents on behalf of @jscottmiller.
2026-09-30 11:53:57 -05:00
Michael Suchacz 1e9dd0abbc feat: add Claude Sonnet 5.5 and GPT-6.1 Sol to Agents and AI Gateway (#30159)
Add Claude Sonnet 5.5 and GPT-6.1 Sol to the Agents known models
catalog, so they appear as suggestions when an admin adds a chat model,
and regenerate the AI Gateway price book from models.dev.

| Model | Added to |
|---|---|
| Claude Sonnet 5.5 | anthropic (`reasoningEffort: high`), azure,
bedrock (`global.anthropic.claude-sonnet-5-5`), openrouter, vercel |
| GPT-6.1 Sol | openai (`reasoningEffort: medium`), azure, openrouter,
vercel |

**Bedrock.** bedrock-runtime serves Sonnet 5.5 only through its global
inference profile, so the catalog curates the `global.` ID without a
bare alias: fantasy's Bedrock client prefixes a bare `anthropic.*` ID
with a regional profile (`us.`, `eu.`, `apac.`) that this model lacks.
bedrock-mantle takes the bare ID and the gateway looks prices up by
exact provider and model, so `overrides.jq` injects the bare ID as a
copy of the global price entry. The two GPT Daybreak alias injections
are gone: models.dev now lists both aliases, and their guards error out
once it does.

**Reasoning effort.** Sonnet 5.5 rejects the `thinking: {"type":
"disabled"}` that fantasy sends for effort `none`, and GPT-6.1 Sol
rejects `none` outright, both with HTTP 400. `applyReasoningEffort` now
maps `none` to `low` for Sonnet 5.5 on the Anthropic and Bedrock
providers, next to the Opus 5.5 clamp, and maps `none` and `minimal` to
`low` for GPT-6.1 Sol on OpenAI and Azure through a new
`chatopenai.IsGPT61Sol`, next to the GPT-6 Astra clamp. GPT-6 Sol
accepts `none` and stays unclamped; the test pins that. AI Gateway's
Bedrock adaptive-thinking allowlist already matched Sonnet 5.5 through
its `anthropic.claude-sonnet-5` substring; a table case now covers it.

**Price book drift.** `prices.json` includes every model of its 8
providers, so the regeneration also brings in upstream changes since the
last refresh: 28 new entries (such as Grok 4.7 on Bedrock and Mistral
Large 2512 on OpenRouter), 7 entries models.dev delisted (mostly free
variants on OpenRouter and Vercel), and 49 price changes (39 OpenRouter,
10 Vercel). Other uncurated upstream models, such as `-pro` and `-fast`
variants and `openai.*` on Bedrock, stay out of the catalog, matching
how earlier releases were curated.

The Reasoning effort section of `coderd/x/chatd/ARCHITECTURE.md`
documents only the GPT-6 Astra clamp and should gain Sonnet 5.5 and
GPT-6.1 Sol. That doc edit is left to a human author.

Validation: regenerating from a fresh models.dev fetch reproduces both
generated files byte for byte. Go tests pass for
`scripts/aibridgepricesgen`, `coderd/aibridge/prices`,
`coderd/x/chatd/chatprovider`, `coderd/x/chatd/chatopenai`, and
`aibridge/intercept/messages`. Removing either clamp with `go test
-overlay` fails the new cases on exactly the Sonnet 5.5 and GPT-6.1 Sol
inputs. Vitest passes for `src/modules/aiModels` (176 tests) and the
Models and Coder Agents settings pages (24 tests). The pre-commit hook
passed; local pre-push checks were skipped, so CI is the first full run.

Review record: no reviews yet. No remote UAT was run; the tested head is
`78abe17e69d`.

> Xum opened this PR on behalf of @ibetitsmike.
2026-09-29 18:46:41 -07:00
Susana Ferreira 34db9ca638 feat: price AI Gateway token usage by provider model (#30090)
## Description

When the requested model has no price, price AI Gateway token usage by
the model the provider reports instead of leaving it unpriced. This
covers providers that resolve an alias to a specific model, such as
`codex-auto-review` served as `gpt-5.6-luna`.

Stacked on #30086, which records the provider model.

Closes https://linear.app/codercom/issue/AIGOV-650

## Changes

- Look up the price for the requested model first, then the provider
model, preferring a custom price over the price book for each model.
- Store the provider model and the model used for pricing
(`priced_model`) on each token usage.
- Add a `provider_model` label to `unpriced_token_usage_records_total`.
- Document the lookup order and the new metric label.

> [!NOTE]
> Generated by Claude Code on behalf of @ssncferreira.
2026-09-29 17:47:28 +01:00
Sas Swart bcb981b839 chore: bump github.com/coder/boundary to v0.11.0 (#30083)
Bumps `github.com/coder/boundary` from v0.10.0 to
[v0.11.0](https://github.com/coder/boundary/releases/tag/v0.11.0). No
transitive dependency changes.

boundary v0.11.0 wraps the `--session-id-inject-target` format and
example in backticks, so:

- The `coder agent-firewall` help golden and
`docs/reference/cli/agent-firewall.md` are regenerated.
- The temporary `<host>`/`<glob>` entry for that page in
`scripts/docshtmlcheck`'s `allowedUnknownTags` is now stale (the
linter's self-clearing guard fails on it), so it is removed. The
allowlist is now passed into `filterAllowed`/`reportFindings` so the
tests use their own fixture instead of depending on the real, now-empty
map.

> Generated by Coder Agents on behalf of @SasSwart.
2026-09-29 17:00:27 +02:00
Thomas Kosiewski 9aa7130e01 feat: evaluate experiment rules per user in discovery and chatd (#29964)
Slice 3 of 4 for runtime experiment rules: evaluate stored rules per
user in discovery, the frontend experiments query, and the pilot
`mcp-tool-search` gate in chatd. There is still no way to edit rules
through the product; that comes in slice 4.

With no rules stored and a healthy database, backend decisions match
today. This PR deliberately changes three things:

1. **Fail closed on rule reads.** If the rules read fails,
`mcp-tool-search` is off, even when it is enabled at startup.
2. **Frontend freshness.** The experiments query now refetches.
3. **Caching.** `/api/v2/experiments` and rendered HTML are sent with
`Cache-Control: no-store`. Static assets keep their long-lived caching.

## What's in it

- **coderd** builds `api.ExperimentEvaluator` from
`experiments.NewDBStore` plus the startup `--experiments` list.
- **Discovery:** `GET /api/v2/experiments` and the `<meta
property="experiments">` tag both use `EnabledExperiments(ctx, caller)`.
- **chatd:** `mcp-tool-search` is now user-scoped.
`chatd.Config.ExperimentEvaluator` is required, so `chatd.New` returns
an error without it. `prepareGeneration` decides once per turn for the
chat owner (`chat.OwnerID`); child chats inherit the owner.
- **Ruleguard** (`userScopedExperimentStaticCheck` in
`scripts/rules.go`): rejects static `Experiments.Enabled(...)` and
`httpmw.RequireExperiment*(...)` checks of user-scoped experiments in
server packages. A codersdk test keeps its list equal to
`codersdk.ExperimentsUserScoped`.
- **Frontend:** the experiments query is keyed by user ID and seeded
from the meta tag only when the embedded user matches. It is stale 60 s
after page load and refetches on mount, window focus and reconnect.
Logout also clears the experiments metadata. The dashboard keeps the
last loaded list when a background refetch fails, instead of replacing
the page with an error.
- **Once per turn:** chatd decides `mcp-tool-search` on a turn's first
step and reuses that decision for the turn's later steps, so a rule
change applies from the next turn and never withdraws a `find_tools`
call the model already made. The decision lives on the chat runner,
keyed by the turn's prompt row, like the stop-nudge tracker. A runner
handoff in the middle of a turn evaluates again.
- **Reference audit:** the only non-test server check of
`mcp-tool-search` is in `generation_preparer.go`, and it now goes
through the evaluator.

## Testing

- **Go:** per-user results and `no-store` on `/api/v2/experiments`; the
per-user meta tag and the HTML cache headers in `site`; group and org
conditions end to end in enterprise; chatd owner eligibility, on → off →
on across turns including a child chat, and a failed rules read.
- **Vitest:** user-keyed seeding, staleness from page load after a
remount, refetch on focus, logout clearing, and the dashboard surviving
a failed refetch.

Remote dogfood UAT through Coder Agents passed on a licensed test
deployment:
- per-user discovery in both the API and the meta tag;
- widening a rule, and turning it off, with no restart;
- `off` beating a startup default, and `inherit` restoring it;
- fail closed on malformed JSON and on an unknown mode, with no
condition text in the logs;
- the `no-store` headers;
- real-model chatd enforcement: `find_tools` offered on turn 1, gone
after `off`, back on turn 3.

That round tested `b57207d4ee`. The later commits change only tests, and
a follow-up round will cover the final head.

**For the author:** `coderd/x/chatd/ARCHITECTURE.md`'s list of the
runner's local state doesn't mention the per-turn trackers (debug turn,
session start, stop nudges, and now this experiment decision). I did not
edit that document.

**Known behavior:** a broken rule logs one error line per evaluation
(per request or turn), with only the experiment, revision, category and
position.

## Stack

1. #29869: evaluator library
2. #29898: storage and audit types
3. **This PR:** enforcement and discovery
4. Control plane (experimental API, `coder exp experiment-rules` CLI,
docs)

> [!NOTE]
> This PR was written by an AI agent (Xum).

---
_Generated with [`xum`](https://github.com/coder/xum) • Model:
`anthropic:claude-opus-5-5` • Thinking: `high`_

---------

Signed-off-by: Thomas Kosiewski <tk@coder.com>
2026-09-29 12:23:00 +00:00
Nick Vigilante d40e24cf7d docs: nest the CLI reference by subcommand (#29956)
## Summary

Nests the CLI reference by subcommand so the Command Line sidebar
collapses from 156 flat entries to 49 top-level verbs.
Each page's URL follows the invocation, its page title is the full
command as inline code, and its sidebar label is only its own verb as
inline code.

| | Before | After |
| --- | --- | --- |
| File | `docs/reference/cli/oauth2-provider_dcr_disable.md` |
`docs/reference/cli/oauth2-provider/dcr/disable.md` |
| URL | `/reference/cli/oauth2-provider_dcr_disable` |
`/reference/cli/oauth2-provider/dcr/disable` |
| Page title (front matter) | `oauth2-provider dcr disable` | `` `coder
oauth2-provider dcr disable` `` |
| Sidebar label (manifest) | `oauth2-provider dcr disable` | ``
`disable` `` |

Resolves DOCS-140 (coder/coder half). Also fixes the Makefile part of
DOCS-1073 / DOCS-912, which this change depends on.

## Changes

- `scripts/clidocgen`: pages nest by command path. A command with
visible subcommands becomes its directory's `index.md`, so
`oauth2-provider/index.md` and `oauth2-provider/dcr/index.md` sit above
`oauth2-provider/dcr/disable.md`. Subcommand table links are relative to
the parent page.
- `scripts/clidocgen`: the manifest "Command Line" section is built from
the command tree (nested, sorted by name, hidden commands skipped)
instead of a flat list. The redundant `coder` child that pointed back at
`reference/cli/index.md` is gone.
- `Makefile`: the CLI rule now replaces `docs/reference/cli/` wholesale,
so stale pages get removed. The manifest rule now runs `clidocgen
-manifest-only` on top of the API staging manifest, so the CLI nav is
generated instead of edited by hand. As a result, the nine pages that
had no nav entry (`ai-gateway*`, `secret enable/disable`, `server
fix-oidc-links`) now have one.
- `scripts/configdocgen` and 48 prose pages: links point at the new
paths (for example `reference/cli/server.md` is now
`reference/cli/server/index.md`).

## How to review/QA this PR

- `docs/reference/cli/**/*` (approximately 190 pages): Each of these
pages are generated automatically. Some pages appear deleted, and some
pages appear new. Every deleted page corresponds to a new page somewhere
else, so no information is being deleted. The docs in this section
should adhere to the following, so a test in the build preview for each
of these items should be satisfactory:
- [ ] Each subcommand of a command gets nested as a child instead of
being displayed as a flat list.
  - [ ] Each parent command is stored in an `index.md` file.
- [ ] The title of each page inside the page should now feature the full
`coder XYZ` command.
- [ ] The title of each page inside the page should now be formatted as
inline code.
- [ ] The title of each page in the sidebar should be the most granular
subcommand of the whole command string (if the page's title is `coder
ai-gateway keys create`, the sidebar label is `create`).
- [ ] The title of each page in the sidebar should now be formatted as
inline code.
  - [ ] The page slug for each page is now the most granular subcommand.
  - [ ] All links now point to the correct page.
- `docs/manifest.json`: The changes here reflect the updated structure
of the `docs/reference/cli` section.
- `docs/**/*.md` outside of `docs/reference/cli/**/*` (approximately 50
pages): The extent of the changes in these pages are just updating the
paths to point to the new page locations.
- [ ] The links on the page all point to the correct new path in
`docs/reference/cli`.
- `scripts/clidocgen` (4 files): This is the Go code responsible for
generating everything inside of `docs/reference/cli/**/*`. A typical
code review here should be satisfactory.
- `scripts/configdocgen` (2 files): these two files just have updates to
the raw `.md` path used by `docs/admin/setup/configuration-reference.md`
and updates the tests to reflect that.
- `Makefile`: This updates the `make` targets for generating the CLI
docs to fit the new structure.

## Merge ordering

Without the renderer PRs deployed, both docs sites show literal
backticks in titles and sidebar labels. Without the redirects, the old
flat URLs return 404.

1. Deploy before merging this PR: coder/docs-engine#176 (beta-docs
renders these titles as inline code) and coder/coder.com#1111 (classic
docs renders them as inline code). Both are safe to ship first, because
titles without backticks render as before.
2. Merge this PR.
3. Merge coder/coder.com#1112 (redirects from the old flat URLs) only
once this content reaches the mainline release branch that serves
unversioned `/docs`.

## Testing

- `go test ./scripts/clidocgen/ ./scripts/configdocgen/
./scripts/docshtmlcheck/`: pass. New tests cover page paths, subcommand
links, the nested manifest, and front matter quoting.
- `make docs/reference/cli/index.md docs/manifest.json
docs/admin/setup/configuration-reference.md`, then a forced rerun: the
second run produced no diff.
- Checked the manifest: all 164 CLI entries resolve to a file, no page
is missing a nav entry, and everything outside "Command Line" is
byte-identical.
- Relative `.md` links across `docs/`: 0 broken, same as `main`.
- `make pre-commit` and `make pre-push`: pass. On earlier pre-push
attempts, different unrelated vitest files timed out under host load.
Those files pass when run alone, and the final push passed with
`PARALLEL_JOBS=1`.

<details>
<summary>Implementation plan and decision log</summary>

**Goal (from DOCS-140 and the requester):** URLs follow the command
hierarchy. Pages are titled with the full command, and sidebar entries
with the last verb only. Both are formatted as inline code.

**Contract with the renderers:** a title wrapped entirely in one pair of
backticks (``^`[^`]+`$``) renders as inline code. Plain-text consumers
(`<title>`, meta, search, llms corpus) strip the backticks. Page title
comes from front matter `title`, and the sidebar label from the manifest
node `title`.

**Decisions**

- A parent command is `dir/index.md` rather than a sibling `dir.md`. The
root CLI page is already `index.md`, and both docs-engine and the legacy
coder.com renderer map `x/index.md` to `/x`.
- The manifest is built from the command tree, not from the written
files, so ordering is deterministic and the nesting matches the pages
exactly.
- The manifest discard bug (DOCS-1073 item 1) is fixed here, because
without that fix the generated nested manifest would never reach the
tree. DOCS-1073 items 2 through 4 (a prose fix and a hidden-command
policy) are out of scope.
- `-manifest-only` keeps the existing parallel `make gen` layout: the
API rule still owns `_gen/manifest-staging.json`, and CLI regeneration
doesn't retrigger swag.
- The renderer and redirect work lives in the companion PRs in
coder/docs-engine and coder/coder.com. It is linked above and not
duplicated here.

</details>

---

> [!NOTE]
> Generated by Coder Agents on behalf of @nickvigilante.
2026-09-28 15:44:10 -04:00
Michael Suchacz 5dde49dffb feat: add chat workspace file upload endpoint (#27079)
## What

Adds `POST /api/v2/chats/{chat}/workspace-files`, which streams the
request body to the chat's bound workspace agent and returns the final
path/name/size, plus the `workspace-file-reference` message part type
that carries the uploaded file's metadata through chat messages:

- `coderd/exp_chats.go`: owner-only middleware (chat RBAC update +
reject archived), workspace exec authorization (403/409 like the
existing chat exec paths), agent selection (prefer the chat's bound
agent, else deterministic `agentselect.FindChatAgent`, without
persisting a binding), 30s dial timeout, chat/ancestor tracking headers
on the agent connection, and agent error passthrough (agent HTTP status
preserved, 502 with a generic detail for transport failures; the real
error is logged). Deleted and non-running workspaces get a specific 409
("Workspace is stopped. Start the workspace before uploading files.")
before agent lookup. Each failed upload logs one structured warning, and
`coderd_chat_workspace_upload_total{status}` counts outcomes. The route
shares the files rate limit and documents its 429.
- Input validation: client-supplied `workspace-file-reference` parts
must point inside `/.coder/chats/<chat-id>/files/` with a matching
basename, non-negative size, and a `workspace_file_workspace_id` equal
to the chat's currently bound workspace (the upload response echoes it
as `workspace_id`). References uploaded before a workspace rebind are
rejected instead of silently sending the agent an unreadable path. The
file name must already be in sanitized form
(`SanitizeWorkspaceUploadName(name) == name`), and the cleaned path and
base media type are what gets persisted; message edits share the same
validation. Parts are rejected on chat creation (no chat ID exists yet).
- `coderd/x/chatd/chatprompt`: references render into the prompt as text
only, `[workspace file: <name> (<size>) at <path>]`; file bytes never
reach the LLM. Title derivation (`TitleText`) uses the bare file name,
so fallback titles carry no internal marker.
- `codersdk`: part-type constants/fields,
`ChatMessageWorkspaceFileReference` builder, and
`Client.UploadChatWorkspaceFile` (RFC 5987 filename in
`Content-Disposition`).

No DB migration: parts persist as JSON in `chat_messages.content`.

## Why

Bridges chat and the workspace filesystem for CODAGT-211: users drop a
file in chat, the bytes land where the agent's tools can operate on
them, and the model learns the path from the reference part.

## How tested

- `go test ./coderd -run 'WorkspaceFile|ChatInput|UploadChat'` including
an integration test that runs a real in-memory agent, uploads through
the coderd route, verifies bytes on disk, and exercises collision
suffixes.
- `go test ./codersdk ./coderd/x/chatd/...`, `make gen`
(swagger/types/docs), `make lint`.
- Rebased onto main 2026-09-21: the chat API moved from
`/api/experimental` to `/api/v2`, so the route is registered in
`coderd/chat_routes.go` and the SDK method lives on `codersdk.Client`.

## Follow-ups

- The files rate limit counts requests, not bytes or concurrency;
whether this route should use a per-user concurrency cap instead, and
whether the new-chat page should keep partial successes on a 429, is
still open
([CODAGT-1171](https://linear.app/codercom/issue/CODAGT-1171)).

Part of CODAGT-211. Stack: agent API -> coderd endpoint (this PR) -> UI.

> This PR was authored by Mux (AI) working on Mike's behalf.
2026-09-26 21:41:25 +02:00
Nick Vigilante 5e12a74f38 fix: run the test recipe under MAKE_TIMED instead of skipping it (#29849) 2026-09-25 08:16:24 -04:00
Lisa Fiander 9700db5b53 chore: update low-risk dependencies (#29860)
## Summary

Updates pinned dependency overrides across three independent pnpm
installs:

| File | Package | From | To |
|------|---------|------|----|
| `package.json` (root) | brace-expansion | 1.1.16 | 1.1.18 |
| `offlinedocs/package.json` | brace-expansion | 1.1.16 | 1.1.18 |
| `scripts/apidocgen/package.json` | form-data | 4.0.4 | 4.0.6 |

All pinned via `pnpm.overrides`; each `pnpm-lock.yaml` is regenerated to
match. There is no `pnpm-workspace.yaml`, so root, `offlinedocs/`, and
`scripts/apidocgen/` are separate installs and each bump is isolated to
its own tooling.

## Testing

Run locally per install:

**Root**
- `pnpm install` — resolves/locks `brace-expansion@1.1.18`; no stale
references
- `pnpm run check-docs` (markdownlint + table formatter, uses js-yaml) —
505 files, 0 errors
- `quicktype` JSON→TS + `biome --version` — pass

**offlinedocs**
- `pnpm install` — resolves/locks `brace-expansion@1.1.18`; no stale
references
- `pnpm run lint:types` / `format:check` / `build` — pass, 453 pages
built

**scripts/apidocgen**
- `pnpm install` — resolves/locks `form-data@4.0.6`; no stale references
- `widdershins --version` + full render of `coderd/apidoc/swagger.json`
(exercises form-data) — rendered successfully

No behavioral changes; dependency version bumps only.
2026-09-25 09:40:45 +01:00
Nick Vigilante 3d30e201b5 docs: move contributor guides out of the published docs into the repo (#29751)
## Merge order

Merge first; then republish the docs; then coder/coder.com#1100 (the
redirects).

## Summary

The nine pages under `docs/about/contributing/` are for people working
in this repository, not for Coder users or administrators, so they leave
the published documentation.
The repository becomes their canonical home.
Until now the root `CONTRIBUTING.md` and `CODE_OF_CONDUCT.md` were
two-line stubs that only linked to the docs site, so contributors on
GitHub were bounced out to read them.

A companion PR in `coder/coder.com` adds redirects from every retired
`/docs/about/contributing/*` URL to the new GitHub location.
It must merge only after this PR has merged and the docs have
republished.

## Changes

| Old path | New path |
| -- | -- |
| `docs/about/contributing/CONTRIBUTING.md` | `CONTRIBUTING.md`
(replaces the stub) |
| `docs/about/contributing/CODE_OF_CONDUCT.md` | `CODE_OF_CONDUCT.md`
(replaces the stub) |
| `docs/about/contributing/SECURITY.md` | deleted; the root
`SECURITY.md` already holds the policy |
| `docs/about/contributing/AI_CONTRIBUTING.md` |
`contributing/AI_CONTRIBUTING.md` |
| `docs/about/contributing/backend.md` | `contributing/backend.md` |
| `docs/about/contributing/frontend.md` | `contributing/frontend.md` |
| `docs/about/contributing/modules.md` | `contributing/modules.md` |
| `docs/about/contributing/templates.md` | `contributing/templates.md` |
| `docs/about/contributing/documentation.md` |
`contributing/documentation.md` (see note below) |

For every moved file: the docs front matter is stripped and its `title`
becomes the H1, relative links are rewritten for the new depth
(`../../../foo` becomes `./foo` at the root and `../foo` under
`contributing/`; the two links into `docs/admin/` become
`../docs/admin/...`), and the docs-only `<div class="tabs">` wrapper in
`CONTRIBUTING.md` is removed.
`CONTRIBUTING.md` gains a short intro that links the code of conduct,
the security policy and the area guides, since GitHub only auto-surfaces
the root files.

`docs/manifest.json` loses the About → Contributing group and
`docs/images/icons/contributing.svg`, which only that group used.

Inbound references updated:

- `README.md`: contribution guide link now points at `CONTRIBUTING.md`
on GitHub.
- `docs/tutorials/example-guide.md` and
`docs/install/releases/feature-stages.md`: link to
`contributing/documentation.md` on GitHub, since a docs page cannot link
to a repo-only file.
- `docs/.style/style-guide/README.md`: the "Relationship to" section
names the new location.
- `scripts/docshtmlcheck/main.go`: both the package comment and the
user-facing error string name `contributing/documentation.md`.
- `scripts/docshtmlcheck/README.md`: relative link to the new location.
- `coderd/database/migrations/migrate_test.go`: `t.Log` link now points
at `contributing/backend.md#database-fixtures-for-testing-migrations`.
- `.github/pull_request_template.md`: AI contribution guidelines link.
- `.github/workflows/contrib.yaml`: commit-message guidelines link in
the PR title check.
- `.github/workflows/test-docs-preview-mapper.sh`: the mapper test case
that used `docs/about/contributing/CONTRIBUTING.md` now uses
`docs/admin/security/audit-logs.md`.
-
`.claude/docs/{ARCHITECTURE,PR_STYLE_GUIDE,WORKFLOWS,DOCS_STYLE_GUIDE}.md`:
links to `CONTRIBUTING.md`.

`git grep about/contributing` is empty after this change.
`CODEOWNERS` had no entry for the old paths.

Note on `documentation.md`: it is docs-authoring guidance and is parked
in `contributing/` for now.
Folding it into `docs/.style/` is tracked separately and is deliberately
not attempted here.

## Testing

- `go run ./scripts/docshtmlcheck` over `docs/`, and again over
`contributing/`, `CONTRIBUTING.md` and `CODE_OF_CONDUCT.md`: clean.
- `go vet ./scripts/docshtmlcheck/...` and `go test
./scripts/docshtmlcheck/...`: 47 tests pass.
- `go vet ./coderd/database/migrations/...`: clean. `go test` for that
package fails to link on this machine (`tailscale.com/tstime/mono:
invalid reference to runtime.nanotime1`) identically on unmodified
`main`, so it is a local toolchain issue and not exercised here; the
change is a string in a `t.Log`.
- `pnpm check-docs` (markdownlint-cli2 and markdown-table-formatter over
`docs/`, `.claude/docs/`, root `*.md`): 495 files, 0 errors. The same
two tools over `contributing/*.md`: 0 errors.
- Every relative link in the eight moved files resolves to an existing
path (scripted check).
- `bash .github/workflows/test-docs-preview-mapper.sh`: the changed
`assert_maps_to` case passes; the six `resolve_ref`/`pages_for_image`
failures it reports are identical on unmodified `main` on macOS.

Linear: [DOCS-975](https://linear.app/codercom/issue/DOCS-975)

---

🤖 Built with AI assistance.
2026-09-24 19:42:18 +00:00
Nick Vigilante d7b6528d7e docs: cap the config reference TOC and drop the Dangerous emoji (#29857)
## Merge order

Standalone; no dependency on the other DOCS-977 PR (task-list
checkboxes).

## Summary

Tracy and Ian flagged the mixed checkbox/emoji styling in the docs
(Slack thread linked from DOCS-977), including the `⚠️ Dangerous`
heading that `scripts/configdocgen` generates from
`codersdk.DeploymentValues`.
The generated configuration reference also has around 130 `h3` headings,
which makes its "On this page" list unreadable.

This PR changes only the generator, `scripts/configdocgen`, and
regenerates `docs/admin/setup/configuration-reference.md` from it.

## Design or Fix

- Emit `toc_depth: 2` frontmatter at the top of the generated page,
before the `<!-- DO NOT EDIT -->` comment.
docs-engine's sync (`packages/docs-build/src/sync.ts`) reads a page's
own frontmatter before stripping it, so this caps the on-page TOC at h2
without any docs-engine change.
- Strip any leading emoji or symbol from generated section headings
(`stripLeadingSymbol`), so the Dangerous heading reads "Dangerous"
instead of "⚠️ Dangerous".
`sectionRank`'s "Dangerous last" ordering still keys off the raw serpent
group name, so ordering is unaffected.
- Render a GitHub `[!CAUTION]` alert directly under the Dangerous
heading, using the group's own codersdk description when it has one,
otherwise a short fallback ("These options can break your deployment or
weaken its security...").
The `⚠️ Dangerous` group name in `codersdk/deployment.go` is unchanged,
because the CLI and other output use it.

## Testing

- `go test ./scripts/configdocgen/...` — 45 passed (7 new/updated cases:
`TestStripLeadingSymbol`, `TestRenderDangerousWithDescription`, and
updated `TestRenderPipeline` assertions for the stripped heading, the
Caution alert, and its position between the heading and the first
option).
- `go vet ./scripts/configdocgen/...` and `gofmt -l
scripts/configdocgen/*.go` — clean.
- `make docs/admin/setup/configuration-reference.md` — regenerated the
page (never hand-edited); this also runs `markdownlint-cli2 --fix` and
`markdown-table-formatter` as part of the make target, both clean.
- `go run ./scripts/docshtmlcheck` — clean.
- Verified with a small Node script against docs-engine's built
`parseFrontmatter`/`extractTitle` (from
`packages/docs-transforms/dist/transform.js`, at the commit that
introduced `toc_depth` support) that a page shaped like the regenerated
one — `toc_depth: 2` frontmatter followed by the `<!-- DO NOT EDIT -->`
comment and then the `# Configuration reference` H1 — parses `toc_depth:
2` and the title correctly.

## QA

Confirmed via the docs-engine sync parser directly (see Testing) that
`toc_depth: 2` is picked up correctly from this file shape.
I did not spin up docs-engine's dev server against this worktree: my
local docs-engine checkout is behind `origin/main` by several commits
including the one that added `toc_depth` support, so a faithful local
preview would first need a docs-engine rebuild against `origin/main`.
Given the direct parser verification above already confirms the
mechanism, I judged the full dev-server rebuild too heavy for the
marginal confidence gained and skipped it.

Expected rendering, based on the diff and docs-engine's documented
`toc_depth` behavior:

- The "On this page" TOC lists only h2 sections (General, Dangerous, and
the rest), not the ~130 option h3s.
- The Dangerous section renders as a plain "Dangerous" heading, followed
immediately by a Caution callout, with no emoji.

## Upstream

No. This is Coder product documentation generation (`codersdk`-driven),
not a Fumadocs engine change.

Linear:
[DOCS-977](https://linear.app/codercom/issue/DOCS-977/docs-replace-emoji-checkboxes-and-emoji-headings-with-one-icon-style)

---
🤖 Built with AI assistance.
2026-09-24 15:21:48 -04:00
Nick Vigilante d5a6047879 docs: restructure Install into the five phase deployment spine (#29304)
Restructures the Install section into the five-phase deployment spine,
so the reading order matches the order a deployment actually happens:
**plan, prepare, install, validate, operate.** Today the first thing a
reader sees is a shell command, and architecture and sizing sit
downstream in Administration, after the point where they would have been
useful.

Resolves [DOCS-869](https://linear.app/codercom/issue/DOCS-869). Part of
[DOCS-685](https://linear.app/codercom/issue/DOCS-685) and
[DOCS-738](https://linear.app/codercom/issue/DOCS-738).

> [!IMPORTANT]
> **Stacked on [#28960](https://github.com/coder/coder/pull/28960)**
(DOCS-871, terminology). Base is that branch, not `main`. Retarget to
`main` after it merges.
>
> **Paired with coder/coder.com
[#1077](https://github.com/coder/coder.com/pull/1077)**, which must
merge **first**. Without it, every inbound link to a moved page 404s in
the gap.

## Deliberately mechanical

No prose changes. The only edits inside moved pages are relative link
and image path repairs, so the diff reads as renames. That is the point:
this is the highest link-breakage change in the project, and it should
be reviewable as such.

The prose work is [DOCS-868](https://linear.app/codercom/issue/DOCS-868)
(phase overview, architecture rewrite, sizing split, actor table) and
[DOCS-872](https://linear.app/codercom/issue/DOCS-872) (new pages).

## The new spine

```text
Install and deploy Coder            /install
├── Coder CLI                                   [unchanged, retired by DOCS-870]
├── 1. Plan your deployment
│   ├── Architecture and components             [from admin/infrastructure]
│   └── Coder Validated Architecture            [from admin/infrastructure]
│       └── 1k / 2k / 3k / 10k users
├── 2. Prepare prerequisites
│   ├── Licensing                               [from admin/licensing]
│   ├── Air-gapped deployments                  [from install/airgap]
│   └── Mirror images and registries            [from install/registry-mirror-artifactory]
├── 3. Install the control plane
│   ├── Docker / Kubernetes / Rancher / OpenShift
│   ├── Cloud marketplace and VM images
│   └── Unofficial install methods
├── 4. Validate your deployment
│   └── Scale testing                           [from admin/infrastructure]
│       ├── Scale test utility
│       └── Scaling best practices
└── 5. Operate and maintain
    ├── External provisioners                   [from admin/provisioners]
    ├── Upgrading (+ best practices, ESR guides)
    └── Uninstall
```

Leaving Install: release channels become `reference/releases`, feature
stages become `reference/feature-stages`.

`docs/admin/infrastructure/` is gone entirely, which is what DOCS-738
recommended.

## Numbers

- 31 pages moved, all as `git mv` renames.
- 1 page deleted: `admin/infrastructure/index.md`, whose content was a
one-line section blurb.
- 5 new placeholder index pages, one per phase, each a sentence of
orientation plus `<children></children>`. The nav schema requires a path
on every entry, so a phase cannot exist without one.
- 265 relative links and asset paths repaired across 57 files.
- `docs/manifest.json` restructured. Entry count goes 460 to 464: plus 5
stubs, minus the deleted Infrastructure entry.

## Reviewer attention, please

**The redirects PR repairs 19 existing redirects, not just adds new
ones.** This was the surprise. `redirects.json` already had entries
whose *destinations* pointed at pages this PR moves, so they would have
started redirecting to 404s. Examples:
`/docs/architecture/architecture`, `/docs/admin/architectures`,
`/docs/scaling/scale-testing`, `/docs/licensing`, `/docs/enterprise`,
`/docs/install/offline`, `/docs/platforms`, `/docs/install/1-click`,
`/docs/admin/upgrade`, `/docs/install/cloud/ec2`, and both
`feature-stages` aliases. All 19 are fixed in the paired PR.

**This PR touches `scripts/`, so it is not docs-only.** Two release
tools write to paths that moved and would have broken silently at
release time:

- `scripts/update-release-calendar.sh`: `DOCS_FILE` now points at
`docs/reference/releases.md`.
- `scripts/releaser/docs.go`: `releasesFile`, and also `kubernetesFile`
and `rancherFile`, which pointed at `docs/install/kubernetes.md` and
`docs/install/rancher.md`.

<details>
<summary>Deliberate scope decisions</summary>

- **Self-managed vs cloud grouping is flattened.** The reconciled IA
nested platform pages under a "Self-managed platforms" parent. That
would need two more placeholder index pages for groups with no content
yet, so the platform pages sit directly under Phase 3 and the grouping
lands with the DOCS-868 prose.
- **Release-notes URLs left alone.** `scripts/releaser/release.go` and
`scripts/release/generate_release_notes.sh` embed
`https://coder.com/docs/install/releases` and `/docs/install/upgrade` in
release-note templates. The redirects cover them, and changing the
strings would require updating Go test fixtures in
`scripts/release/main_internal_test.go`. Better as its own change.
- **In-product doc links left alone.** Several `site/src` pages link to
`coder.com/docs/admin/provisioners` and similar. Those are absolute
URLs, so the redirects cover them. Updating them is a frontend change
that would pull frontend review into a docs restructure.
- **`admin/security/audit-logs.md` line 237** is hand-written prose
below the generated table, and its `../licensing/index.md` link had to
move with the licensing page. Verified `make gen` does not revert it.
- **`reference/feature-stages.md`** has generated blocks whose relative
links changed depth with the move. The generator named in its own marker
comment (`scripts/release/docs_update_feature_stages.sh`) no longer
exists in the repo, so those links were repaired by hand and the file is
effectively hand-maintained. Worth its own issue.
- **`install/cli.md` keeps its nav slot** here. DOCS-870 retires it.

</details>

## Verification

| Check | Result |
|---|---|
| Broken relative link scan | 0 in `install/`, `admin/`, `reference/`.
21 remain, all pre-existing illustrative examples in
`docs/.style/style-guide/**` and one malformed URL in `docs/README.md` |
| `make gen` | clean, manifest not clobbered, phase structure intact |
| `make lint/markdown` | pass, 508 files, 0 errors |
| `make lint/emdash` | pass |
| `make lint/docs-html` | pass |
| `make lint/prose` | pass, 1 intentional demo error, no new findings |
| `make pre-commit` | pass, including `lint/go`, `lint/ts`,
`lint/shellcheck`, and a slim build |
| Manifest audit | every path resolves to a file on disk; entry-by-entry
diff shows only the expected moves |

Generated with Coder Agents on behalf of @nickvigilante.


<details>
<summary>CI note: doc-check failure is infrastructure, not this
PR</summary>

`Analyze PR for Documentation Updates Needed` fails before it reviews
anything: the Coder account behind `agents-chat-action` returns `User is
not active (status = "suspended")`. It isn't a required check.
`check-docs` (linkspector) and every other check pass on `58ed9e3e4b`.

_CI note added by Coder Agents on behalf of @nickvigilante._
</details>
2026-09-24 13:51:38 -04:00
Nick Vigilante eb893f536d docs: move section page to index.md (#29858) 2026-09-24 13:13:27 -04:00
Ehab Younes bacb4fc6ed feat: expose per-app session counts (#29351)
Track individual applications across deployment statistics, Prometheus, telemetry, and the dashboard instead of exposing only fixed family totals. Preserve existing fields and gauges so current consumers can migrate without breaking.

- Add per-app counts with normalized identifiers and registry-backed display names, families, and bundled icons. Unrecognized apps retain their identifiers.
- Limit each statistics report to 64 distinct app names, combining excess names into an explicit overflow bucket.
- Group dashboard connections by family, including an Other category, with alphabetical per-app tooltips and distinct loading and empty states.
- Export app-keyed Prometheus counts separately from current family metadata, allowing queries to enrich historical counts without rewriting stored samples.
- Document family queries with unknown fallback, deployment isolation, and replica handling, alongside collection behavior and migration guidance.
- Include per-app counts in telemetry, update API schemas and documentation, and add regression coverage across collection, metrics, and dashboard behavior.

Closes #27411
2026-09-24 15:12:28 +03:00
Michael Suchacz 5b5432ce73 feat: add GPT-6 Sol, GPT-6 Luna, and Claude Opus 5.5 to Agents and AI Gateway (#29762)
Add the models released on 2026-09-22 (GPT-6 Sol, GPT-6 Luna, and Claude
Opus 5.5) plus Grok 4.7 to the Agents known models catalog, so they
appear as suggestions when an admin adds a chat model in `/agents`
settings, and pick up their prices in the AI Gateway price book. chatd
and AI Gateway's Bedrock path also stop producing Opus 5.5 thinking
shapes it rejects.

| Model | Added to |
|---|---|
| GPT-6 Sol, GPT-6 Luna | openai, azure, openrouter, vercel |
| Claude Opus 5.5 | anthropic, azure, bedrock (plus `global.` alias),
openrouter, vercel |
| Grok 4.7 | openrouter (`x-ai/grok-4.7`), vercel (`spacexai/grok-4.7`)
|
| Qwen3.8 Max | vercel (`alibaba/qwen3.8-max-0902`; already curated on
openrouter) |

Every model is already on models.dev, so `overrides.jq` is unchanged.
Sol and Luna use `reasoningEffort: medium` like GPT-6 Astra and the
GPT-5.6 models. Opus 5.5 uses `high` like every other Opus, Fable, and
Mythos entry, and joins the Anthropic thinking-mode pin in
`knownModelsGenerated.test.ts`. Anthropic documents `medium` as the Opus
5.5 default, but the catalog value seeds both the default and the max
effort in the admin form (`applyKnownModelDefaults.ts`), so `medium`
would also stop users from picking `high`. Gateway entries follow the
existing convention of no reasoning default and upstream display names;
Qwen3.8 Max reuses its openrouter display name. New entries sort ahead
of the models they supersede.

**chatd.** Opus 5.5 always runs adaptive thinking and rejects `thinking:
{"type": "disabled"}` with HTTP 400. The pinned fantasy fork sends
exactly that for the `none` effort on Claude 5+ models (it exempts only
Mythos and Fable), and the effort picker offers `none` whenever a model
config sets a max effort. `applyReasoningEffort` now maps `none` to
`low` for Opus 5.5 on the Anthropic and Bedrock providers, next to the
existing GPT-6 Astra clamp. The new `ClampsNoneForClaudeOpus55` subtest
fails without the fix. The Reasoning effort section of
`coderd/x/chatd/ARCHITECTURE.md` documents the Astra clamp and should
gain a line for this one; that doc edit is left to a human author.

**AI Gateway on Bedrock.** Opus 5.5 also rejects `thinking: {"type":
"enabled", "budget_tokens": N}`. On the Bedrock InvokeModel path,
`augmentRequestForBedrockInvokeModel` in
`aibridge/intercept/messages/base.go` rewrote adaptive thinking into
that legacy shape for every model outside its adaptive-only allowlist.
`bedrockModelRequiresAdaptiveThinking` now includes Opus 5.5, so
adaptive requests pass through, legacy requests (Claude Code falls back
to them behind the gateway) convert to adaptive, and
`output_config.effort` is kept. `output_config.format` is still
stripped, as for the other adaptive-only models; before this change the
whole `output_config` was dropped. Two new
`TestAugmentRequestForBedrock_AdaptiveThinking` cases fail without the
fix.

Deliberately not added: the `-pro` (openrouter) and `-fast` (vercel)
variants, matching how Astra and the GPT-5.6 models are curated;
`openai.*` models on Bedrock; regional Bedrock profiles; `~` rolling
aliases; `copilot`; `azure/grok-4.6`, which models.dev lists without
pricing; vendors not curated on the gateways (Xiaomi MiMo V2.6, Nex
N2.5, and others); and variants of curated models (GLM 5.3 FlashX,
Qwen3.8 Omni Flash). All of these that have prices are still in the
price book.

Known gaps, unchanged by this PR:

- The price book stores one flat price per model, so the long-context
tier for Sol and Luna (above 272K input tokens) is not applied, as for
Astra.
- The Bedrock adaptive-only allowlist still lacks the already curated
Opus 5, Fable 5, and Fable 5.1, plus Mythos 5 and 5.1, which also reject
`budget_tokens`, so their adaptive requests still take the legacy
conversion. They predate this PR and need the same allowlist entry.
- AI Gateway forwards a client's explicit `thinking: {"type":
"disabled"}` unchanged on every path, so Opus 5.5 answers it with its
documented 400, as the direct API does.

`knownModelsGenerated.json` and
`coderd/aibridge/prices/data/prices.json` are regenerated with `make
gen/aibridge-prices` from one models.dev snapshot. Besides the rows
above, both carry the upstream drift the weekly scheduled refresh would
produce. The catalog has nine repriced gateway rows, including vercel
`openai/gpt-5.6-sol`, which moves from $2/$10 to $4/$20 per 1M
input/output tokens and now matches OpenAI's direct price. `prices.json`
gains 41 rows (26 for the models above across every listed provider, 15
for other new upstream models), drops 4 that upstream delisted, and
reprices 30.

Validation: `go test ./scripts/aibridgepricesgen/...
./coderd/aibridge/prices/... ./coderd/x/chatd/chatprovider/...
./aibridge/...`, `pnpm test src/modules/aiModels/` (176 tests),
`golangci-lint` on `chatprovider` and `aibridge/intercept/messages`, and
`biome check` on `knownModels/` pass.

Review log (4 completed review passes):

- Codex round 1 on `03decfde855`: two P1 findings. Bedrock adaptive
thinking for Opus 5.5 was confirmed by a failing test and fixed. The
request to add a TODO to `coderd/x/chatd/ARCHITECTURE.md` was declined;
the chatd paragraph above flags that section for a human.
- Codex round 2 on `0156b3e5997`: one P2 finding, that the chatd clamp
misses opaque Bedrock application inference profile ARNs and fantasy
then sends disabled thinking. Refuted: fantasy omits the thinking field
when it cannot parse a Claude version from the model ID, so Opus 5.5
keeps its default adaptive thinking.
- Codex round 3 on `0156b3e5997`: two P1 findings. The claim that
`minimal` reaches Opus 5.5 verbatim was refuted: fantasy normalizes it
to `low`. The request to rewrite a client's explicit `disabled` thinking
for Opus 5.5 on Bedrock was declined: the gateway forwards it unchanged
on every path, before and after this PR (see Known gaps).
- Codex round 4 on `0156b3e5997`: no findings ("Didn't find any major
issues").

> Xum opened this PR on behalf of @ibetitsmike.
2026-09-23 14:10:24 +02:00