skills/mine/spec-peer-review/SKILL.md
Run one requested external review of an approved spec, design doc, RFC, or detailed PRD; produce findings for user-selected incorporation. Excludes implementation reviews and automatic or looping approval gates.
npx skillsauth add pedronauck/skills spec-peer-reviewInstall 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.
Review a saved design only when the user requests an independent review. Honor existing approval to review it as-is. Incorporate only selected findings; additional rounds remain opt-in.
qa/ directory for .compozy/tasks/<slug>/, otherwise .peer-reviews/<UTC-timestamp>/.<out>/peer-review-findings-roundN.md. Preserve prior rounds.Read the spec and its linked contract index. Include relevant requirements, concrete examples, test contracts, ADR decisions, and the instructions for affected surfaces. Expand into other context when a dependency or contradiction needs it; do not load every memory file.
Use the applicable markers in references/quality-markers.md; missing contracts become findings, not an automatic stop before a requested review.
Read references/peer-review-prompt.md and fill its placeholders, including exact findings path, round, resolved runtime/model, and context paths. The reviewer may write only that findings file.
Launch through the available independent reviewer runtime (a configured native reviewer, Compozy session, or herdr TUI). Use that runtime's documented commands; do not assume a retired compozy exec API. Preserve existing worker identity for follow-up rounds when available.
Inspect the findings, cited evidence, and changed paths. Validate the artifact:
bash <skill-dir>/scripts/validate-findings.sh --kind techspec --round <N> --path <out>/peer-review-findings-roundN.md
Report the verdict, material findings, limitations, and artifact path. Missing/malformed output is an incomplete review, never READY. Repair an operational failure within the authorized round; do not start additional substantive rounds automatically.
Apply only user-selected findings and record the disposition with changed paths. Reuse authorization already given; ask only for a selection that is still missing. On an explicitly requested next round, send changed files, prior findings/disposition, and the new output path rather than the unchanged corpus again.
This skill does not authorize commits, publishing, or approving a design. A real product/security/compatibility constraint can block readiness; missing ritual sections and stylistic preferences cannot.
testing
Assess DDD fit; discover domains with EventStorming, define language and bounded contexts, and design aggregates, value objects, events, and repositories. Excludes architecture-only audits, product specs, and CQRS/Event Sourcing catalogs.
development
Build terminal UIs with ratatui following 2026 Rust best practices. Use when: (1) Creating new TUI apps, (2) Adding widgets/layouts, (3) Keyboard navigation/state management, (4) Image integration via ratatui-image, (5) Async event handling, (6) Release optimization. Covers v0.30.0+ API, Elm Architecture, StatefulWidget, color-eyre.
development
Find 10x product opportunities and high-leverage improvements. Use when user wants strategic product thinking, mentions '10x', wants to find high-impact features, or says 'what would make this 10x better', 'product strategy', or 'what should we build next'.
development
You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.