[Birfy]

Fixes #609

Tip SHA: 4c8d98c

Reproduce

The rule text says risk:T* is set by the controller from the diff, never by hand, and the
mechanism does not hold up: github-actions[bot] labelled #574 at 07:28:43Z, while #606, #610,
#611, #612, #616, #617, #618 and #619 carry no risk:T* at all even though risk tier (report only) succeeded on every one of them. The owner decision on the issue is ②: keep the workflow
read-only, never label — and make the computed tier auditable instead of remembered.

At this tip, node .github/scripts/risk-gate.mjs --pr <n> ends its report with

risk-tier: T3; basis: AGENTS.md=T3, plugins/social/src/social.js=T3

so a review can quote the tier the rules produced instead of restating a number.

What changed

• .github/scripts/risk-gate.mjs: judge() records every changed path (and a rename's old path)
  with the tier tierOf() gave it, and summaryLine() renders the final line of the report from
  that same result — highest tier first, paths capped at 10 with a +N more tail. render() ends
  the job summary with it, after the prose, so the two can never disagree.
• .github/scripts/risk-gate.test.mjs: four cases — the line's exact shape for a T3 PR, the
  ordering and the cap, the no-changed-files case, and "the line is the last line of the report".
• AGENTS.md gains a bullet under the label section: quote the summary line, the workflow never
  writes a label, the controller copies it on its poll, no review transcribes a tier.
• .github/labels.json: every risk:T* description keeps advisory and the merge gate recomputes it and now also names the controller as the writer (all ≤ 100 characters).
• docs/controller.md: the tier stays auditable without the label — quote the summary line.
• .github/workflows/risk-tier.yml: header comment only — no permission added, no step changed.
  It keeps contents: read and pull-requests: read and still never labels, comments or merges.

What I ran

• TZ=UTC node --test .github/scripts/*.test.mjs at this tip (with git on PATH) →
  144/144.
• node --test .github/scripts/risk-gate.test.mjs → 18/18, including the four new cases.
• node .github/scripts/risk-gate.mjs behaviour checked by hand through the harness result:
  judge([AGENTS.md, plugins/social/src/social.js]) → risk-tier: T3; basis: AGENTS.md=T3, plugins/social/src/social.js=T3; no files → risk-tier: T0; basis: none (no changed files).
• changelog.mjs check not run: a change users cannot see needs no fragment (changelog.d/README).

Red without the source change

• Removing the basis.push([name, t]) line from judge() (the tests untouched) →
  risk-gate.test.mjs 16/18, 2 red, exactly summaryLine: one machine-readable line… and
  summaryLine: highest tier first….
• Removing summaryLine(result) from the end of render() instead → 17/18, 1 red, exactly
  render: the machine-readable line is the last line of the report. Restored → 18/18.

Not run

• The workflow itself (it needs Actions): the change to risk-tier.yml is a comment, and the
  existing the workflow stays report only test still passes with it.
• npm run verify end to end: no core/node_modules in this sandbox (the same on main).
• The risk-tier job on this PR runs the base copy of the script, so its summary will only show
  the new line after merge — that is the workflow's normal base-copy behaviour, not a gap here.

Risk

T3 by the repository's own rules (.github/**, AGENTS.md): a reporting script plus text. It
adds no permission, no network call, no label, comment or merge, and cannot block anything — the
job is still continue-on-error: true and not a required check. The only behaviour it changes is
the last line of that job's summary. The one thing a reader must not conclude is that the label
now arrives on time: it does not, and the docs keep saying why (#629).

[Birfy]

Verdict: PASS at 4c8d98c
Written-by: Reviewer

Reviewed this tip only. Nine checks completed/success, mergeable clean, no earlier conclusion here. Risk tier T3 (.github/**, AGENTS.md), so it needs a maintainer's approve as well; I have not queued it.

What I checked

• The line and the prose come from one judge() call. basis is filled where each path's tier is decided, deduplicated by path and sorted by tier then path, and render appends summaryLine(result) as the last line, so the two cannot disagree. SUMMARY_LIMIT caps the list at 10 with +N more.
• For this PR's own files the gate gives T3, 131 changed lines, 6 files, none generated, and the line reads .github/labels.json=T3, .github/scripts/risk-gate.mjs=T3, .github/scripts/risk-gate.test.mjs=T3, .github/workflows/risk-tier.yml=T3, AGENTS.md=T3, docs/controller.md=T0.
• The claim the labels now make is true. controller.mjs:24-27,129 adds/removes exactly risk:T0..T3 from the gate's tier, so "the controller labels it, the merge gate recomputes it" matches the code; risk-tier.yml itself only writes the job summary (its change here is a comment).
• labels.json: 13 labels before and after, and the only content changes are the four risk:T* descriptions — agent-task, hold and the rest are untouched.
• Tests: node --test ".github/scripts/*.test.mjs" → 153/153; risk-gate.test.mjs + risk-label-docs.test.mjs → 21/21. Red without the source change: .github/scripts/risk-gate.mjs reverted to 8005fc9d, tests untouched → risk-gate.test.mjs cannot even load (summaryLine missing), 0 pass. Restored → 153/153.
• A rename is listed under both names with its own tier (docs/x.md <- core/src/host/x.js → core/src/host/x.js=T2, docs/x.md=T0), so the highest tier keeps a matching basis entry.

Non-blocking

• A tier raised without a path has no basis entry. Size and an incomplete file list raise the tier in why but add nothing to basis, so the line can contradict itself: a 13-file, 325-line core/src/host/** change prints risk-tier: T3; basis: core/src/host/p0.js=T2, … — every listed path is T2 while the tier is T3, and only the prose Why says why. The limits make this the common case (T2: 250 lines / 12 files), not a corner. Adding one synthetic entry (for example (size)=T3 / (file list)=T3) when a non-path rule bumps the tier would make the machine line answer "what set it" on its own; the AGENTS.md sentence "the tier and the paths that set it" reads as if it always does.
• A path containing , or = breaks the line's parse. basis: docs/a,b=c.md=T0 cannot be split reliably; either escape the two characters or say in the docs that the line is human-readable, not a parse contract.
• The labels.json reformat (one object per line → pretty-printed) turns a four-value edit into a whole-file diff. Harmless, but it hides the real change from anyone reading the diff.

No other finding; the line is report-only, adds no token scope, and no review is expected to copy a tier onto the PR — the controller does that on its next poll.