skills/review-prd/SKILL.md
Review and verify that a Ralph PRD's user stories have been implemented correctly. Reads prd.json, steps through each user story's acceptance criteria, runs tests, checks code, and uses browser verification. Produces a pass/gap/untestable summary. Triggers on: review prd, verify prd, check ralph requirements, review requirements, are the stories done, verify user stories, prd review.
npx skillsauth add richardwu/agent-skills review-prdInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Verify that each user story in a Ralph prd.json has been correctly implemented by checking acceptance criteria, running tests, and inspecting code.
prd.json (check ralph skill directory or current project root)CLAUDE.md for testing instructions (test commands, lint commands, dev server setup)prd.json.branchName and confirm you're on it (or the work has been merged)Before per-story review, run project-level checks once to get baseline status:
tsc --noEmit, pnpm typecheck)Record results — these apply to all stories with "Typecheck passes" or "Tests pass" criteria.
For each story in userStories, evaluate every acceptance criterion:
Code-verifiable criteria (e.g., "Add status column", "Install package X"):
Test-verifiable criteria (e.g., "Typecheck passes", "Tests pass"):
Browser-verifiable criteria (e.g., "Verify in browser using dev-browser skill"):
Behavioral criteria (e.g., "Clicking delete shows confirmation dialog", "Filter persists in URL params"):
For each criterion, record one of:
After reviewing all stories, output a structured summary:
## PRD Review: [project name]
Branch: [branchName]
Date: [today]
### Passed
Stories where ALL acceptance criteria passed.
- **US-XXX: [title]** — All N criteria passed
### Gaps Found
Stories where one or more criteria have gaps or non-standard behavior.
- **US-XXX: [title]**
- [criterion text] — [what's wrong or non-standard]
- [criterion text] — [what's wrong or non-standard]
### Untestable
Stories (or specific criteria) that couldn't be verified.
- **US-XXX: [title]**
- [criterion text] — [why it couldn't be tested]
### Repo-wide Check Results
- Typecheck: PASS/FAIL [details if fail]
- Lint: PASS/FAIL/SKIPPED
- Tests: PASS/FAIL/SKIPPED [X passed, Y failed]
"passes": true — verify them anyway, Ralph may have been wrongdevelopment
Convert PRDs to prd.json format for the Ralph autonomous agent system. Use when you have an existing PRD and need to convert it to Ralph's JSON format. Triggers on: convert this prd, turn this into ralph format, create prd.json from this, ralph json.
documentation
Generate a Product Requirements Document (PRD) for a new feature. Use when planning a feature, starting a new project, or when asked to create a PRD. Triggers on: create a prd, write prd for, plan this feature, requirements for, spec out.
development
Multi-agent code review using 3 parallel subagents (Claude, GPT, Gemini) to review branch changes against origin/main. Validates findings against actual diffs and posts consolidated report to GitHub PR. Use before opening a PR, when wanting multiple AI perspectives on code changes, or for large changesets that might truncate in single-agent reviews. Triggers on "multi-agent review", "review my branch", "review changes against main", "AI code review", or "review before PR".
development
Fix issues on the current PR: address bot (eg Claude Code, CodeRabbit, or custom GHA) review comments and fix failing CI checks. Use when asked to fix PR, fix review comments, fix CI, or fix checks. Triggers on: fix pr, fix review, fix ci, fix checks, fix failing checks.