trove-review
Code review skill that checks for common issues, security vulnerabilities, and style violations. Use when reviewing code changes or PRs.
- License 许可
- MIT
- 所属仓库
- bravew/trove
- 最近更新
- 2026-11-12
- Stars
- ★ 10
- Forks
- 1
01 前言 Front Matter
- name
- trove-review
- metadata
- { "trove-category": "pull-requests", "trove-category-label": "Pull requests" }
- description
- Code review skill that checks for common issues, security vulnerabilities, and style violations. Use when reviewing code changes or PRs.
03 SKILL.md 正文
<!-- AUTO-GENERATED from SKILL.md.tmpl — do not edit directly --> <!-- Regenerate: bun run build:skills -->
Session Init
This skill ships Trove conventions. Prefer existing project patterns over generic best practices when they conflict.
Code Review Guidelines
Review Checklist
When reviewing code, check for these categories in order:
If a PRP or implementation plan drove the change, read it first and review against its requirements. With no explicit scope, review staged changes; with a PR number, inspect the PR metadata and files; with paths, read those files in full before judging.
1. Correctness
- Does the code do what it claims to do?
- Are edge cases handled (null, empty, boundary values)?
- Are error paths handled correctly?
2. Security
- No hardcoded secrets, API keys, or credentials
- Input validation on all external data
- SQL injection prevention (parameterized queries)
- XSS prevention (proper escaping/sanitization)
- Authentication/authorization checks in place
3. Performance
- No N+1 queries
- Appropriate use of caching
- No unnecessary re-renders (React) or re-computations
- Pagination for large datasets
4. Maintainability
- Clear naming (variables, functions, files)
- Reasonable function length (under 50 lines preferred)
- No code duplication
- Appropriate abstraction level
5. Testing
- Tests cover the happy path
- Tests cover error/edge cases
- Mocks are used appropriately
- Test names describe the behavior being tested
6. Reader load
- Layers to trace — flag one-caller wrappers and indirection that exists only to forward a call; collapse them.
- State to hold — flag mutable scope wider than it needs to be.
- The <30s test: can a new reader answer "where does X come from?" and "what can change X?" without spelunking? If not, say where it breaks.
Deep mode (adversarial, optional)
On request ("tear this apart", "stress test", "find blind spots") or for sensitive surfaces, run a deeper pass: review the change through several independent lenses (correctness, security, failure modes, maintainability), optionally via read-only sub-agents on disjoint concerns. State the change's intent first so the review challenges the execution, not the goal. Synthesize one verdict and sort every finding into Act on / Consider / Noted / Dismissed. Do not silently rewrite the diff — the deliverable is the verdict. For a true multi-model cloud review, point the user at /code-review ultra.
Delegated exploration
When the diff is wide enough that linear reading is wasteful, dispatch read-only sub-agents in parallel for bounded investigation. Examples of good delegations:
- "Map every callsite of the modified function in
services/. Return file:line." - "Verify whether the new validation path is reachable from the public API surface. Read-only. Yes/no plus one cite."
- "Check the test files touched in this diff for missing edge-case coverage (empty input, boundary values). Return findings or 'no gaps'."
Each delegation should declare scope, expected output shape, and a depth budget. Don't delegate "review the whole diff" — that's the parent's job. Anti-patterns and host-specific examples are in docs/orchestration.md.
For sensitive surfaces (auth, payments, secret handling, untrusted input),
also run trove-security-review and trove-secret-scan — those are this
skill's benefits-from.
Review Output Format
Structure your review as:
## Summary
One-paragraph overview of the changes.
## Issues Found
### 🔴 Critical (must fix)
- [file:line] Description
### 🟡 Important (should fix)
- [file:line] Description
### 🔵 Suggestion (nice to have)
- [file:line] Description
## Positive Observations
- Things done well worth calling out
04 同仓库更多 Skills
using-trove
Skill discipline anchor for the Trove marketplace. Loaded once per session; routes the agent to the right trove-* skill before code, planning, review, or git actions.
trove-write-skill
Skill authoring workflow. Use when creating or updating a Codex or Trove skill, including rubrics, tasks, metadata, and validation.
trove-worktree
Git worktree workflow for spikes, risky experiments, and broad refactors that should not disturb the user's current branch.
trove-why
Reconstruct design rationale and decision history from evidence — source control, issue tracker (Linear), docs (Notion), chat (Slack), error tracking (Sentry) — with calibrated confidence. Use for "why does X work this way", "why did we pick Y", regressions, and postmortems. Use trove-explain for how it works now.
trove-verify
Completion verification workflow. Use before claiming done, fixed, complete, ready, or shipped.
trove-unslop
Cut AI tells from writing — PR descriptions, release notes, specs, commit bodies, and replies. Use before finalizing prose, or when the user asks to clean up, de-slop, or make text sound human.