pr-review

PR Review

Run a structured multi-agent review of a GitHub pull request and present a synthesized verdict.

PR: $ARGUMENTS (if non-empty, use as PR number; if empty, auto-detect from current branch).

Invocation

/pr-review           # reviews the PR for the current branch
/pr-review {{number}}  # reviews a specific PR by number

Natural language triggers: "review this PR", "review PR 123", "review my pull request".

What This Skill Does

The pipeline fetches PR metadata and the full diff, then runs a Haiku triage agent to classify changed files by review priority. Four Sonnet agents run in parallel — each examining the diff from a different angle: project convention compliance, bug scanning, git history context, and historical PR feedback on the same files. An Opus agent then cross-validates their findings and issues a verdict of APPROVE or REQUEST CHANGES. The main agent presents the synthesis output and keeps all prior agent outputs available for follow-up questions.

Protocol

Before doing anything else, read ${CLAUDE_SKILL_DIR}/references/protocol.md in full. It defines every stage of the pipeline: exact commands, verbatim prompt templates for each subagent, and failure handling for every error scenario. Do not proceed without reading it.

Constraints

• Do not post the review as a GitHub comment unless the user explicitly requests it
• Keep all prior agent outputs in context so the user can ask follow-up questions
• If a stage fails, follow the failure handling rules in ${CLAUDE_SKILL_DIR}/references/protocol.md exactly
• No conversational text during execution — only tool calls until the final summary