Orchestrate a specialist panel to review pull requests and produce a ship recommendation.
npx skills add https://github.com/microsoft/apm --skill apm-review-panel
Please help me install this Agent Skill. Skill: apm-review-panel Source: https://github.com/microsoft/apm/tree/main/.apm/skills/apm-review-panel Command: npx skills add https://github.com/microsoft/apm --skill apm-review-panel
This skill orchestrates a cross-discipline advisory review of a labeled pull request in microsoft/apm. It runs a panel of five mandatory specialists plus a test-coverage specialist (active on every PR that touches src/) plus two conditional specialists (auth, doc-writer), all running in their own agent threads, and a CEO synthesizer. The orchestrator is the sole writer to the PR: ONE recommendation comment, no verdict labels, no merge gating. The panel surfaces findings, prioritizes follow-ups, and renders a ship recommendation; it is advisory—human review remains essential.
Trigger an advisory pass on a PR by running the apm-review-panel against the touched diff to generate a single advisory comment.
High-intent search queries and answers about installing and using this skill.
Automated cross-functional pull request reviews are handled by orchestrating a panel of specialists across architectural, UX, security, and growth surfaces to surface findings for human weighing. This multi-agent advisory approach generates a single ship recommendation comment without automated merge gating.
A multi-persona advisory review panel is a multi-agent workflow where five mandatory specialists, conditional experts, and a CEO synthesizer run in separate threads. It produces a single prioritized recommendation comment and structured findings overview instead of a pass/fail verdict.
No, the advisory review panel does not automatically block or gate pull request merges. The orchestrator outputs a single recommendation comment with no verdict labels, meaning human review and decision-making remain essential before shipping.
Yes, you can run conditional specialists like auth and doc-writer on code pull requests. A test-coverage specialist also activates automatically on any PR that touches the src/ directory, alongside the fixed roster of architectural and growth panelists.
A multi-agent review panel should be used for non-trivial pull requests that touch architecture, UX, security, or growth surfaces. It is designed to surface findings across multiple lenses for human weighing, whereas standard reviews may miss cross-disciplinary impacts.
The best way to generate a single ship recommendation is using a CEO synthesizer agent to consolidate inputs from parallel specialist threads. This orchestrator writes one unified advisory comment to the PR, prioritizing follow-ups and rendering a clear ship recommendation for human reviewers.
jpoleyGenerate secure patches for code vulnerabilities and output unified diffs.
pedroscostaFetch unresolved GitHub PR comments, apply feasible fixes, and resolve threads.
inbharataiAnalyzes code for bugs, security vulnerabilities, performance bottlenecks, and style violations.
daudaudinangReview LP spec packages for vagueness, gaps, and edge-case risks.
