poweruser-feature-audit

Audits large feature PRs by comparing external API research against implementation behavior.

19.6k|2.6k|Updated Jun 21, 2024
One-click install
npx skills add https://github.com/pydantic/pydantic-ai --skill poweruser-feature-audit
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: poweruser-feature-audit
Source: https://github.com/pydantic/pydantic-ai/tree/main/.agents/skills/poweruser-feature-audit
Command: npx skills add https://github.com/pydantic/pydantic-ai --skill poweruser-feature-audit

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Reviewing a large feature PR (new provider API, new modality, new subsystem) by reading the diff first anchors your expectations to the implementation, so you end up reviewing the code against itself. This Skill enforces an unbiased audit: research the underlying APIs and integrator experience from external sources first, design the ideal test suite from a power-user perspective, and only then gap-compare it against the PR to produce evidence-backed review items.

Core Features & Use Cases

  • Bias-gated phased workflow: Scoping, parallel external research, ideal test-suite design, gap analysis, test execution, and verification run in strict order, with the PR diff unread until the design phase is complete.
  • Source-linked research fan-out: Parallel subagents research each provider/API/protocol plus practitioner lessons, with every claim carrying a specific documentation link and GA/beta/deprecated status.
  • Verified, precedent-backed review items: Findings are classified into covered/partial/missing/unimplemented buckets, re-verified against the PR HEAD, and drafted into inline comments with minimal fixes, runnable proof commands, and mandatory precedent links.
  • Use Case: A PR adds a realtime voice API integration. The skill researches the provider's protocol docs and community pain points, designs the ideal behavioral test catalog, then identifies which behaviors the PR leaves untested or unimplemented, producing draft review comments the user triages before anything is posted.

Quick Start

Run the poweruser-feature-audit skill on this feature PR URL to produce an independent, evidence-backed audit with drafted review items.

Frequently Asked Questions about poweruser-feature-audit

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I audit a large feature PR without bias?

Research the underlying APIs and protocols from external sources first, design an ideal test suite from that knowledge, and only then open the PR diff for gap comparison. This ordering prevents the implementation from anchoring your expectations.

How to review a PR that adds a new provider API integration?

Scope the PR with a fact sheet, research each provider's documentation and practitioner pain points in parallel, then classify every ideal behavior as covered, partial, missing, or unimplemented. Verify each finding against the PR HEAD before drafting review comments.

Does this skill post review comments to GitHub automatically?

No. The skill ends at drafted review items and inline comment drafts stored as local files. Posting is a separate human-gated step, and drafts must be re-verified against the PR's current HEAD before posting.

When should I not use a power-user feature audit?

Do not use it for general line-by-line diff review, code style checks, or completing narrow fixes. It targets behavioral completeness and operational robustness of large feature PRs that wrap external APIs or protocols.

What artifacts does the feature audit produce?

It writes a scope fact sheet, per-provider research files, an ideal test-suite design, per-area gap analyses, verified review items, and draft PR comments under local-notes/<feature>-audit/, each resumable across sessions.