chore(claude): rename pr skill to pull-request and fix automation gaps#3160
Merged
PierreBrisorgueil merged 6 commits intomasterfrom Feb 24, 2026
Merged
chore(claude): rename pr skill to pull-request and fix automation gaps#3160PierreBrisorgueil merged 6 commits intomasterfrom
PierreBrisorgueil merged 6 commits intomasterfrom
Conversation
- Rename .claude/skills/pr/ → .claude/skills/pull-request/ with shorter description so the skill loads reliably in system-reminder - Fix monitor loop: push before reply/resolve, CI retry fallback, thread ID source - Add preliminary review pass after gh pr ready (CodeRabbit and ready-triggered bots) - Add 10-iteration safety limit clarification (preliminary pass not counted) - Add guardrail in CLAUDE.md to always invoke /pull-request when shipping work - Update feature/SKILL.md to reference /pull-request Closes #3159
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review infoConfiguration used: defaults Review profile: CHILL Plan: Pro 📒 Files selected for processing (1)
🚧 Files skipped from review as they are similar to previous changes (1)
📝 WalkthroughWalkthroughRenames the PR skill from Changes
Sequence Diagram(s)sequenceDiagram
participant Dev as "Developer"
participant Skill as "pull-request Skill"
participant GH as "GitHub (gh + API)"
participant CI as "CI"
participant Repo as "Repository"
Dev->>Skill: /pull-request (create draft PR)
Skill->>Repo: git push branch (create draft PR)
Skill->>GH: gh pr create (draft)
Skill->>CI: wait for CI to start (poll)
CI-->>Skill: CI queued / running / result
Skill->>GH: list unresolved review threads (GraphQL)
GH-->>Skill: return threads + IDs
Skill->>Skill: evaluate actionable comments
alt actionable comments exist
Skill->>Repo: apply fixes (batch commits) and push
Skill->>GH: gh pr ready / update
Skill->>CI: wait for CI (loop)
else no actionable comments and CI green
Skill->>Dev: stop (CI green + zero actionable)
end
Estimated code review effort🎯 4 (Complex) | ⏱️ ~45 minutes Possibly related PRs
Suggested labels
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In @.claude/skills/pull-request/SKILL.md:
- Around line 157-162: The TOTAL calculation uses gh api --paginate piped to jq
'length', which only returns the last page's length; update the TOTAL assignment
(the shell variable TOTAL in the snippet that calls gh api
repos/.../issues/.../comments --paginate) to slurp and aggregate all pages
before computing length (use jq's slurp mode to merge the per-page arrays and
compute the combined length), so the script sums counts across pages rather than
reading only the final page.
ℹ️ Review info
Configuration used: defaults
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (6)
.claude/skills/feature/SKILL.md.claude/skills/pr/SKILL.md.claude/skills/pr/references/monitoring.md.claude/skills/pull-request/SKILL.md.claude/skills/pull-request/references/monitoring.mdCLAUDE.md
💤 Files with no reviewable changes (2)
- .claude/skills/pr/SKILL.md
- .claude/skills/pr/references/monitoring.md
There was a problem hiding this comment.
Pull request overview
Renames the embedded Claude PR workflow skill from /pr to /pull-request and updates the skill content to close gaps in the autonomous PR monitoring loop (CI waiting, feedback handling, and stop conditions) within the repo’s internal .claude/ tooling.
Changes:
- Renamed the PR skill to
pull-requestand added updated monitoring/reference documentation under.claude/skills/pull-request/. - Updated
CLAUDE.mdand the/featureskill to point contributors toward/pull-requestfor PR lifecycle operations. - Removed the legacy
.claude/skills/pr/skill docs.
Reviewed changes
Copilot reviewed 6 out of 6 changed files in this pull request and generated 5 comments.
Show a summary per file
| File | Description |
|---|---|
CLAUDE.md |
Updates guardrails and the skills table to reference /pull-request. |
.claude/skills/pull-request/SKILL.md |
New PR lifecycle skill playbook with updated autonomous monitor loop steps. |
.claude/skills/pull-request/references/monitoring.md |
New gh/GraphQL reference commands for CI and review thread monitoring. |
.claude/skills/pr/SKILL.md |
Removes the old /pr skill doc. |
.claude/skills/pr/references/monitoring.md |
Removes the old monitoring reference doc. |
.claude/skills/feature/SKILL.md |
Adds a step directing users to /pull-request once verify passes. |
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Nitpick comments (1)
.claude/skills/pull-request/references/monitoring.md (1)
56-66:--paginate | jq 'map(...)'produces separate arrays per page.
gh api --paginateemits one JSON array per page;jq 'map(...)'(without-s) processes each independently, outputting multiple arrays. For display this is functional, but inconsistent with the-s 'add | …'pattern already applied in SKILL.md. Consider using slurp mode for a single flat list:♻️ Suggested consistency fix
-gh api repos/$OWNER/$REPO/pulls/$PR/comments --paginate \ - | jq 'map({id, user: .user.login, body: .body[0:100], line})' +gh api repos/$OWNER/$REPO/pulls/$PR/comments --paginate \ + | jq -s 'add | map({id, user: .user.login, body: .body[0:100], line})'-gh api repos/$OWNER/$REPO/issues/$PR/comments --paginate \ - | jq 'map({id, user: .user.login, body: .body[0:100]})' +gh api repos/$OWNER/$REPO/issues/$PR/comments --paginate \ + | jq -s 'add | map({id, user: .user.login, body: .body[0:100]})'🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In @.claude/skills/pull-request/references/monitoring.md around lines 56 - 66, The current gh api commands using --paginate piped to jq 'map(...)' produce separate arrays per page; change the jq usage to slurp mode and merge pages into a single array (use jq -s 'add | map({id, user: .user.login, body: .body[0:100], line})' for the pull-request comments command and jq -s 'add | map({id, user: .user.login, body: .body[0:100]})' for the issue comments command) so the output is a single flattened list while keeping the same fields.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In @.claude/skills/pull-request/SKILL.md:
- Around line 139-157: Replace the literal "<number>" placeholder with the
previously defined PR variable ($PR) in the bash blocks that call the GitHub CLI
(specifically the commands like "gh pr checks <number> --watch" and the retry
loop using "gh pr checks <number>") so the script uses the PR variable; update
the occurrences in sections 6a and 6f to reference $PR (matching how other
blocks use $PR) and ensure all instances of "<number>" in those blocks are
changed to $PR so the commands execute with the actual PR number.
---
Nitpick comments:
In @.claude/skills/pull-request/references/monitoring.md:
- Around line 56-66: The current gh api commands using --paginate piped to jq
'map(...)' produce separate arrays per page; change the jq usage to slurp mode
and merge pages into a single array (use jq -s 'add | map({id, user:
.user.login, body: .body[0:100], line})' for the pull-request comments command
and jq -s 'add | map({id, user: .user.login, body: .body[0:100]})' for the issue
comments command) so the output is a single flattened list while keeping the
same fields.
ℹ️ Review info
Configuration used: defaults
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (7)
.claude/skills/pull-request/SKILL.md.claude/skills/pull-request/references/monitoring.md.github/copilot-instructions.md.github/prompts/pr.prompt.md.github/prompts/pull-request.prompt.mdAGENTS.mdREADME.md
💤 Files with no reviewable changes (1)
- .github/prompts/pr.prompt.md
There was a problem hiding this comment.
♻️ Duplicate comments (1)
.claude/skills/pull-request/references/monitoring.md (1)
28-29:⚠️ Potential issue | 🟡 Minor
monitoring.mdstill has the unconditional--watch— inconsistent withSKILL.md.The retry loop here has no
CHECKS_FOUNDguard, so line 29'sgh pr checks "$PR" --watchstill executes unconditionally after all 5 retries, which can hang indefinitely if CI never starts.SKILL.md(lines 146–164) was updated with the correct guard; this reference snippet was not.🛠️ Proposed fix — align with
SKILL.md+CHECKS_FOUND=0 for i in 1 2 3 4 5; do if output=$(gh pr checks "$PR" 2>&1); then if echo "$output" | grep -q "no checks reported"; then sleep 30 # checks not started yet else - echo "$output" && break # checks detected + echo "$output" && CHECKS_FOUND=1 && break # checks detected fi else echo "$output" >&2 && sleep 30 # gh command failed, retry fi done -gh pr checks "$PR" --watch +[ "$CHECKS_FOUND" -eq 1 ] && gh pr checks "$PR" --watch🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In @.claude/skills/pull-request/references/monitoring.md around lines 28 - 29, The snippet in monitoring.md runs gh pr checks "$PR" --watch unconditionally; replicate the SKILL.md fix by guarding the final watch invocation with the same CHECKS_FOUND check used in the retry loop: only run gh pr checks "$PR" --watch if CHECKS_FOUND is true (i.e., wrap the command in the same conditional used in SKILL.md), ensuring the retry loop's CHECKS_FOUND variable controls whether the watch is executed to avoid hanging when CI never starts.
🧹 Nitpick comments (1)
.claude/skills/pull-request/references/monitoring.md (1)
86-92: Use GraphQL variables instead of a hard-coded string literal forTHREAD_ID.Single quotes prevent shell variable expansion, so
"THREAD_ID"in the mutation is treated as a literal string. The idiomaticgh api graphqlapproach is to pass the value as a GraphQL variable using an additional field flag (e.g.,-f threadId="$THREAD_ID"), which avoids quoting issues entirely.♻️ Proposed refactor — use GraphQL variable binding
+THREAD_ID="<id-from-list-query>" gh api graphql -f query='mutation { - resolveReviewThread(input: {threadId: "THREAD_ID"}) { + resolveReviewThread(input: {threadId: $threadId}) { thread { isResolved } } -}' +}' -f threadId="$THREAD_ID"Or equivalently, declare the variable in the operation signature:
THREAD_ID="<id-from-list-query>" gh api graphql \ -f threadId="$THREAD_ID" \ -f query='mutation Resolve($threadId: ID!) { resolveReviewThread(input: {threadId: $threadId}) { thread { isResolved } } }'🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In @.claude/skills/pull-request/references/monitoring.md around lines 86 - 92, Replace the hard-coded string literal "THREAD_ID" in the GraphQL mutation passed to gh api graphql with a GraphQL variable binding: declare a variable (e.g., threadId) in the mutation signature and pass its value via gh's -f flag (for example -f threadId="$THREAD_ID"), then use that variable in the resolveReviewThread input (input: {threadId: $threadId}) so shell variable expansion works correctly and avoids the single-quote quoting issue.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Duplicate comments:
In @.claude/skills/pull-request/references/monitoring.md:
- Around line 28-29: The snippet in monitoring.md runs gh pr checks "$PR"
--watch unconditionally; replicate the SKILL.md fix by guarding the final watch
invocation with the same CHECKS_FOUND check used in the retry loop: only run gh
pr checks "$PR" --watch if CHECKS_FOUND is true (i.e., wrap the command in the
same conditional used in SKILL.md), ensuring the retry loop's CHECKS_FOUND
variable controls whether the watch is executed to avoid hanging when CI never
starts.
---
Nitpick comments:
In @.claude/skills/pull-request/references/monitoring.md:
- Around line 86-92: Replace the hard-coded string literal "THREAD_ID" in the
GraphQL mutation passed to gh api graphql with a GraphQL variable binding:
declare a variable (e.g., threadId) in the mutation signature and pass its value
via gh's -f flag (for example -f threadId="$THREAD_ID"), then use that variable
in the resolveReviewThread input (input: {threadId: $threadId}) so shell
variable expansion works correctly and avoids the single-quote quoting issue.
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #3160 +/- ##
=======================================
Coverage 89.86% 89.86%
=======================================
Files 52 52
Lines 1164 1164
Branches 234 234
=======================================
Hits 1046 1046
Misses 107 107
Partials 11 11 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
This was referenced Feb 24, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
.claude/skills/pr/→.claude/skills/pull-request/and fix multiple automation gaps in the monitor loopprskill was not being loaded into Claude Code's system-reminder (description too long + late addition), causing the PR workflow to be silently skipped. Several loop gaps also caused incorrect behavior when the skill was used.Scope
.claude/skills/(internal tooling only, no application code)nonelowValidation
npm run lintnpm testGuardrails check
.env*,secrets/**, keys, tokens)Notes for reviewers
.claude/andCLAUDE.mdSummary by CodeRabbit
Documentation
Chores