This skill should be used when the user asks to "review this PR", "check the pull request", "review PR
From pr-toolsnpx claudepluginhub aaronbassett/agent-foundry --plugin pr-toolsThis skill uses the workspace's default tool permissions.
Dispatches parallel agents to independently tackle 2+ tasks like separate test failures or subsystems without shared state or dependencies.
Executes pre-written implementation plans: critically reviews, follows bite-sized steps exactly, runs verifications, tracks progress with checkpoints, uses git worktrees, stops on blockers.
Guides idea refinement into designs: explores context, asks questions one-by-one, proposes approaches, presents sections for approval, writes/review specs before coding.
Orchestrate a comprehensive pull request review by launching multiple specialist agents in parallel.
When this skill is invoked, you should:
/pr-tools:review-pr - Current branch PR/pr-tools:review-pr --pr 123 - Specific PR in current repo/pr-tools:review-pr --repo user/repo --pr 123 - Any repo PR/pr-tools:review-pr --dry-run - Save to /tmp instead of posting/pr-tools:review-pr --pr 123 --dry-run - Dry run for specific PRpython3 /tmp/cpr.py pr-tools to get PLUGIN_ROOTUse gh pr view "$PR_NUM" --repo "$REPO" --json number,title,state,isDraft,author,body,files,additions,deletions to get:
Launch the pr-tools:pr-validator agent with the PR data to:
If PR is not reviewable:
Using a SINGLE Task tool call with multiple invocations, launch all these agents in parallel:
For each detected language:
TODO Finder:
SDD Task Verifier (conditional):
Wait for all agents to complete and collect their outputs.
If some agents fail or timeout:
Launch the pr-tools:pr-commenter agent with:
The agent will:
After all agents complete, present a clean summary:
If dry-run:
Review complete. Saved to /tmp/pr-review-{number}.md
Summary:
- {count} critical issues
- {count} important improvements
- {count} suggestions
[if SDD detected: - {verified}/{total} P1 tasks verified]
If posted:
Review posted successfully.
View PR: https://github.com/{repo}/pull/{number}
Summary:
- {count} critical issues
- {count} important improvements
- {count} suggestions
[if SDD detected: - {verified}/{total} P1 tasks verified]
Handle these common errors gracefully:
gh auth logingh pr view with --json for structured data (easier to parse)gh pr diff to get diff content for agentsgh api for finding existing commentsWhen launching agents, use a single Task tool call with multiple invocations:
<invoke name="Task">
<parameter name="subagent_type">pr-tools:code-reviewer</parameter>
<parameter name="description">Review Python code</parameter>
<parameter name="prompt">Review PR #123 in user/repo for Python code...</parameter>
</invoke>
<invoke name="Task">
<parameter name="subagent_type">pr-tools:code-reviewer</parameter>
<parameter name="description">Review TypeScript code</parameter>
<parameter name="prompt">Review PR #123 in user/repo for TypeScript code...</parameter>
</invoke>
<invoke name="Task">
<parameter name="subagent_type">pr-tools:todo-finder</parameter>
<parameter name="description">Find TODO comments</parameter>
<parameter name="prompt">Find TODOs in PR #123...</parameter>
</invoke>
This ensures all agents run concurrently for optimal performance.