agentleFS
nuncaeslupusnuncaeslupus/literary-critic/.claude/skills/review/SKILL.md

review

Skill0 starsChanged 6 days agoMITView on GitHub

Reviews a PR, diff, design doc or proposal for risks and standards. Use when the user wants something reviewed. Not for release sign-off (ship).

192 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use close-spec, daily-arxiv and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • dzhngSkill

    close-spec

    When a spec under specs/<feature>/ is done shipping (a write-spec build, or any planned task that produced a spec), archive it to specs/done/ and rewrite it from a build-plan into a durable rationale — the why, the principles, the invariants — pointing back to the real code for the how. Use when implementation has landed and the plan no longer matches what shipped, or the user says a feature/spec is finished. Pairs with [write-spec](../write-spec/SKILL.md) (the plan this closes) and [review](../review/SKILL.md) (run before closing).

    Also in 5 projects that use this

  • LambenthanSkill

    daily-arxiv

    每日从 arXiv 拉取新论文,过滤相关性,auto-ingest 高优先级论文,检测 SOTA 更新

    Also in 5 projects that use this

  • dzhngSkill

    design-with-images

    Explore UI and visual design with image generation, then iterate the real implementation against the user-selected concept. Use when brainstorming visual directions, offering mockup variants, or implementing an approved generated design; do not stop at the mockups or the first code approximation.

    Also in 5 projects that use this

  • LambenthanSkill

    exp-eval

    实验判决门:Review LLM 独立评判实验结果 → 4 种判决路径 → 自动更新 claims confidence、ideas status、graph edges

    Also in 5 projects that use this

  • +5 more from Lambenthan
  • +3 more from dzhng

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 156 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read PR description/ticket.

    line 21
  2. Distinctive:

    If the intention is unclear, ask before reviewing code.

    line 22
  3. Distinctive:

    If engineering standards exist in the host repo (a project-level engineering-standards skill or equivalent), verify:

    line 26
  4. Distinctive:

    Do tests actually assert meaningful behavior (not just "it doesn't crash")?

    line 48
  5. Distinctive:

    If no tests exist and the change is non-trivial → flag as blocker

    line 49
  6. Very specific:

    Do added tests earn their cost?

    line 50

“Common” is measured: kept by at least 6 in 10 of the 156 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Review Workflow

CANARY: review-loaded-2026-05-19-da60b2aa44b817de

Reads status/specification.md (the statement of intent) and audits the diff against it, surfacing drift between spec and implementation.

Record every finding with a severity and a confidence, including low ones; deciding which findings block or get dropped is a separate step at the verdict, so nothing is filtered out before it is seen.

Steps

Step 1: Understand intention

Read PR description/ticket. State the intention in one sentence before proceeding. If the intention is unclear, ask before reviewing code.

Step 2: Check engineering standards compliance

If engineering standards exist in the host repo (a project-level engineering-standards skill or equivalent), verify:

  • Code structure follows repo conventions
  • Tech stack follows the approved stack
  • Code conventions (naming, async patterns, type hints, no hardcoded secrets)
  • API versioning (backwards compatible? New version if breaking?)
  • Security (auth on new endpoints, input validation, audit logging, no sensitive data in logs)
  • Configuration (URLs via env vars, safe defaults)

Step 3: Check functional correctness

  • Does the code actually solve the stated problem?
  • Are there edge cases not handled?
  • Are error paths handled correctly (not silently swallowed)?
  • Is the data flow correct (inputs → processing → outputs)?
  • Are there race conditions, deadlocks, or concurrency issues?

Step 4: Check test coverage

  • Are new code paths covered by tests?
  • Are edge cases tested?
  • Are error conditions tested?
  • Do tests actually assert meaningful behavior (not just "it doesn't crash")?
  • If no tests exist and the change is non-trivial → flag as blocker
  • Do added tests earn their cost? Flag ones that grep prose, re-run a measurement a gate already runs, pin constants or glue, duplicate another test, or rebuild an expensive fixture per test — load claude-arsenal:core:execution § references/test-cost.md when the diff adds tests.

Gate evidence: for a plan with task Gates, every Evidence log row must be complete and meet its gate — the gate-check skill's run_gate.py --input status/plan.md answers it.

Exit 1 (a gate failed or lacks evidence) is a blocker. Exit 2 means no Gate column or a usage error: confirm the plan path first, then treat it as a should-flag only for a plan that predates gates.

Hard blocker rule: production code changes with zero test companions in the diff are Request Changes — except config-only, docs-only, or refactor with existing green tests covering the touched paths. When waiving on the refactor exception, the reviewer states the exception applied and asserts the green-test evidence explicitly (CI link or local test run).

Step 5: Check operational readiness

  • Will this change affect deployment? (migration, feature flag, coordination)
  • Are there observability gaps? (new endpoints without tracing, new errors without alerts)
  • Is the change backwards compatible with running instances during deploy?
  • Is rollback straightforward?

Step 6: Filter and produce the output

Now sort the findings: blockers, recommendations, and anything dropped as a false positive (say why). Load references/template.md when writing the review document.


Abbreviation

Abbreviated review = Step 1 + quick diff scan + verdict, where the host repo's CLAUDE.md allows it.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.