refactor(ci): review pull requests with one senior-engineer role

The review job ran the official code-review plugin, which fans a pull
request out to five Sonnet reviewers plus a Haiku scorer per finding and
drops everything scored under 80, and the briefing file spent most of its
lines overriding that plugin. Both are gone: the job hands one Senior
Software Engineer prompt to the action inline, the way the issue analyst
does, and denies the Agent tool so the single role is mechanical rather
than a request.

REVIEW.md moves from the emoji markers to CRITICAL/HIGH/MEDIUM/LOW with a
pre-existing qualifier. The uncapped rule is scoped to findings the pull
request introduced or worsened so it cannot collide with the cap of three
pre-existing ones. "A finding is a report, not a patch" stays as it was.

Workflow housekeeping: GH_TOKEN, REPO and PR live in the job env instead
of six step copies; the skip gate is per pull request, so a head pushed
after the automatic review is reviewed only on @claude review; the comment
counters sum gh's per-page jq output, which read "0\n0" as a review on a
pull request with more than 100 comments; --max-turns rises to 300 because
every read now costs the single agent a turn instead of a subagent.
This commit is contained in:
Sanaei
2026-09-08 18:46:30 +02:00
parent 47d2303334
commit acf3603dc8
4 changed files with 146 additions and 178 deletions
+101 -59
View File
@@ -33,6 +33,10 @@ jobs:
pull-requests: write
issues: read
id-token: write
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
REPO: ${{ github.repository }}
PR: ${{ github.event.pull_request.number || github.event.issue.number }}
steps:
- name: Record when this run started
id: started
@@ -43,8 +47,6 @@ jobs:
if: github.event_name == 'issue_comment'
continue-on-error: true
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
REPO: ${{ github.repository }}
COMMENT_ID: ${{ github.event.comment.id }}
run: gh api "repos/${REPO}/issues/comments/${COMMENT_ID}/reactions" -f content=eyes
- uses: actions/checkout@v7
@@ -55,9 +57,6 @@ jobs:
- name: Pin the head this run reviews
id: pinned-sha
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
REPO: ${{ github.repository }}
PR: ${{ github.event.pull_request.number || github.event.issue.number }}
PAYLOAD_SHA: ${{ github.event.pull_request.head.sha }}
COMMENT_AT: ${{ github.event.comment.created_at }}
run: |
@@ -80,23 +79,19 @@ jobs:
exit 1
fi
echo "sha=${HEAD_SHA}" >> "$GITHUB_OUTPUT"
# An automatic re-review of a head that already has one spends a whole run
# to reach the same conclusion, so settle it here rather than in the model.
- name: Skip a head that already has a review
# One automatic review per pull request: a later push is reviewed only
# when a maintainer asks for it with `@claude review`.
- name: Skip a pull request that already has a review
id: reviewed
if: github.event_name == 'pull_request_target'
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
REPO: ${{ github.repository }}
PR: ${{ github.event.pull_request.number }}
HEAD_SHA: ${{ steps.pinned-sha.outputs.sha }}
run: |
set -euo pipefail
posted=$(gh api "repos/${REPO}/issues/${PR}/comments" --paginate \
--jq "[.[] | select(.user.login == \"github-actions[bot]\") | select((.body | contains(\"Reviewed head:\")) and (.body | contains(\"${HEAD_SHA}\")))] | length")
--jq '[.[] | select(.user.login == "github-actions[bot]") | select(.body | contains("Reviewed head:"))] | length' \
| awk '{n += $1} END {print n + 0}')
if [ "$posted" != "0" ]; then
echo "done=true" >> "$GITHUB_OUTPUT"
echo "::notice::#${PR} already carries a review of ${HEAD_SHA}; nothing to review."
echo "::notice::#${PR} already carries a review; nothing to review."
fi
# Read-only, and pinned to one immutable commit: this job holds a
# write-scoped token, so running anything out of pr-head/ would be a pwn-request.
@@ -107,26 +102,6 @@ jobs:
path: pr-head
persist-credentials: false
allow-unsafe-pr-checkout: true
# The skill reads CLAUDE.md on its own but not REVIEW.md, and knows nothing
# of pr-head/ or this run's head: the brief is the only way both reach it.
- name: Brief the reviewer
if: steps.reviewed.outputs.done != 'true'
env:
REPO: ${{ github.repository }}
PR: ${{ github.event.pull_request.number || github.event.issue.number }}
HEAD_SHA: ${{ steps.pinned-sha.outputs.sha }}
TRIGGER: ${{ github.event_name }} / ${{ github.event.action }}
run: |
set -euo pipefail
{
cat .github/claude/review-job.md
printf '\n## This run\n\n'
printf -- '- Repository: %s\n' "$REPO"
printf -- '- Pull request: #%s\n' "$PR"
printf -- '- Head under review, checked out read-only in pr-head/: %s\n' "$HEAD_SHA"
printf -- '- Trigger: %s\n' "$TRIGGER"
printf -- '- CI on that head: gh api repos/%s/commits/%s/check-runs\n' "$REPO" "$HEAD_SHA"
} > "$RUNNER_TEMP/review-brief.md"
- uses: anthropics/claude-code-action@v1
id: review
if: steps.reviewed.outputs.done != 'true'
@@ -137,25 +112,99 @@ jobs:
github_token: ${{ secrets.GITHUB_TOKEN }}
claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
allowed_non_write_users: "*"
plugin_marketplaces: "https://github.com/anthropics/claude-code.git"
plugins: "code-review@claude-code-plugins"
prompt: "/code-review:code-review max --comment ${{ github.repository }}/pull/${{ github.event.pull_request.number || github.event.issue.number }}"
# allowedTools only pre-approves; it denies nothing. Only the deny
# list stops the review executing what it just checked out.
# Claude Code loads a CLAUDE.md or .claude/rules/ file the moment a file
# beside it is read, so a fork's copy under pr-head/ would brief its own review.
settings: '{"claudeMdExcludes": ["**/pr-head/**"]}'
# allowedTools only pre-approves; it denies nothing. Only the deny list
# stops the review executing what it just checked out, or delegating.
claude_args: |
--model claude-opus-5
--effort xhigh
--max-turns 100
--allowedTools "mcp__github_inline_comment__create_inline_comment,Bash(gh api:*),Bash(gh pr diff:*),Bash(grep:*),Bash(rg:*),Bash(ls:*),Bash(find:*),Bash(sed:*),Bash(git log:*),Bash(git show:*),Bash(git diff:*),Bash(go doc:*),Bash(go env:*),Read,Glob,Grep,WebFetch,WebSearch"
--disallowedTools "Bash(go build:*),Bash(go run:*),Bash(go test:*),Bash(go generate:*),Bash(go install:*),Bash(make:*),Bash(npm:*),Bash(npx:*),Bash(pnpm:*),Bash(yarn:*),Bash(node:*),Bash(bash:*),Bash(sh:*),Bash(docker:*),Bash(chmod:*),Edit,Write,NotebookEdit"
--append-system-prompt-file ${{ runner.temp }}/review-brief.md
--max-turns 300
--allowedTools "mcp__github_inline_comment__create_inline_comment,Bash(gh api:*),Bash(gh pr view:*),Bash(gh pr diff:*),Bash(gh pr comment ${{ env.PR }}:*),Bash(grep:*),Bash(rg:*),Bash(ls:*),Bash(find:*),Bash(sed:*),Bash(git log:*),Bash(git show:*),Bash(git diff:*),Bash(git blame:*),Bash(go doc:*),Bash(go env:*),Read,Glob,Grep,WebFetch,WebSearch"
--disallowedTools "Agent,Bash(go build:*),Bash(go run:*),Bash(go test:*),Bash(go generate:*),Bash(go install:*),Bash(make:*),Bash(npm:*),Bash(npx:*),Bash(pnpm:*),Bash(yarn:*),Bash(node:*),Bash(bash:*),Bash(sh:*),Bash(docker:*),Bash(chmod:*),Edit,Write,NotebookEdit"
prompt: |
You are a Senior Software Engineer performing a production-grade code
review of pull request #${{ env.PR }} in ${{ env.REPO }}. You are the
only reviewer: no other role, no subagent, no second pass. What you
post is the whole review.
Your goal is to identify real defects and meaningful risks, not to
criticise style or suggest refactoring nobody needs. Review the entire
change in the context of the existing codebase, not the hunks alone.
Prioritise, in this order:
1. Correctness
2. Bugs and edge cases
3. Security
4. Concurrency and race conditions
5. Performance
6. Data integrity
7. API and backward compatibility
8. Error handling
9. Maintainability
10. Test coverage
Report only what is actionable and supported by evidence from the
code. Do not invent hypothetical problems. Do not nitpick formatting
or personal style. Do not request tests merely to raise coverage.
If the implementation is correct, say so. Do not manufacture findings.
For every finding, explain the problem, why it can happen, which code
is affected (`file:line`), and the impact. Mark it with one severity:
CRITICAL - security, data loss, corruption, or severe production failure
HIGH - a significant functional or production issue
MEDIUM - a real bug or a meaningful reliability or performance problem
LOW - a minor but legitimate issue
THE RUBRIC
Read `REVIEW.md` at the repository root before the diff, and follow it:
what is HIGH in this repository, the checks to always run, what not to
report, the verification bar, the volume cap and the shape of the
comment. It also settles the one thing a finding never carries: the
fix. Name where the fix belongs, never what it is - no patch, no
snippet, no suggestion block, no rewrite in prose. The maintainer
decides the change.
WHAT IS CHECKED OUT WHERE
The working tree is the BASE branch. The head under review,
${{ steps.pinned-sha.outputs.sha }}, is checked out read-only in
`pr-head/`: read and grep the changed files there, and treat anything
outside it as the pre-merge baseline. Never build, install or execute
anything from `pr-head/`. This job holds a write-scoped token, and
running pull-request code with it is the workflow vulnerability
`REVIEW.md` calls blocking.
CI IS THE BUILD
You cannot build or test here, but CI already ran on the head. Read
its check runs with
`gh api repos/${{ env.REPO }}/commits/${{ steps.pinned-sha.outputs.sha }}/check-runs`
and report what they concluded instead of writing that verification
was unavailable. A required check that failed, or never ran on this
head, is itself a finding.
ROUNDS
Trigger: ${{ github.event_name }} / ${{ github.event.action }}. On an
`@claude review`, review in full even when an earlier comment of yours
exists, focusing on the commits since the head it names, and apply the
rounds rule in `REVIEW.md`: after the first review of a pull request,
MEDIUM and above only.
THE COMMENT
This run ends the moment you end your turn, and a run that ends
without posting has failed. Anchor each finding to its line with an
inline comment, then post the summary with
`gh pr comment ${{ env.PR }} --repo ${{ env.REPO }}`. The summary opens
with the tally, carries the line
`Reviewed head: ${{ steps.pinned-sha.outputs.sha }}`, and ends with the
coverage list `REVIEW.md` asks for, whether or not you found anything.
- name: Upload the run transcript
if: always()
env:
NODE_OPTIONS: ""
uses: actions/upload-artifact@v7
with:
name: claude-review-${{ github.event.pull_request.number || github.event.issue.number }}-${{ github.run_id }}-${{ github.run_attempt }}
name: claude-review-${{ env.PR }}-${{ github.run_id }}-${{ github.run_attempt }}
path: ${{ runner.temp }}/claude-execution-output.json
if-no-files-found: ignore
retention-days: 7
@@ -165,9 +214,6 @@ jobs:
id: throttled
if: ${{ !cancelled() && steps.review.outcome == 'failure' }}
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
REPO: ${{ github.repository }}
PR: ${{ github.event.pull_request.number || github.event.issue.number }}
TRANSCRIPT: ${{ runner.temp }}/claude-execution-output.json
run: |
set -euo pipefail
@@ -182,23 +228,19 @@ jobs:
echo "skipped=true" >> "$GITHUB_OUTPUT"
echo "::notice::No review of #${PR}: ${reason}."
gh pr comment "$PR" --repo "$REPO" --body "No review ran on this head: ${reason}. Nothing in this pull request was examined. A maintainer can ask for one with \`@claude review\`."
# updated_at, not created_at: a re-review may edit its earlier comment.
# --paginate prints one jq count per page, so the pages are summed.
- name: Fail if the review posted nothing
if: ${{ !cancelled() && steps.pinned-sha.outcome == 'success' && steps.reviewed.outputs.done != 'true' && steps.throttled.outputs.skipped != 'true' }}
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
REPO: ${{ github.repository }}
PR: ${{ github.event.pull_request.number || github.event.issue.number }}
HEAD_SHA: ${{ steps.pinned-sha.outputs.sha }}
STARTED_AT: ${{ steps.started.outputs.at }}
run: |
set -euo pipefail
head=$(gh api "repos/${REPO}/pulls/${PR}" --jq '.head.sha')
# updated_at, not created_at: the skill may update its existing sticky comment.
# "Reviewed head:" as well as the SHA — the bot's other comments quote SHAs too.
posted=$(gh api "repos/${REPO}/issues/${PR}/comments" --paginate \
--jq "[.[] | select(.user.login == \"github-actions[bot]\") | select((.updated_at >= \"${STARTED_AT}\") or ((.body | contains(\"Reviewed head:\")) and (.body | contains(\"${head}\"))))] | length")
inline=$(gh api "repos/${REPO}/pulls/${PR}/comments" --paginate \
--jq "[.[] | select(.user.login == \"github-actions[bot]\") | select(.updated_at >= \"${STARTED_AT}\")] | length")
since="[.[] | select(.user.login == \"github-actions[bot]\") | select(.updated_at >= \"${STARTED_AT}\")] | length"
posted=$(gh api "repos/${REPO}/issues/${PR}/comments" --paginate --jq "$since" | awk '{n += $1} END {print n + 0}')
inline=$(gh api "repos/${REPO}/pulls/${PR}/comments" --paginate --jq "$since" | awk '{n += $1} END {print n + 0}')
if [ "$posted" = "0" ] && [ "$inline" = "0" ]; then
echo "::error::The review run ended without posting a review of ${head} on #${PR}. Read the uploaded transcript before re-running."
echo "::error::The review run ended without posting a review of ${HEAD_SHA} on #${PR}. Read the uploaded transcript before re-running."
exit 1
fi