agentleFS
lalulalilalulali/vespyr/.agents/skills/review/SKILL.md

review

Skill11 starsChanged 11 days agoMITView on GitHub

Standalone code review on current changes — outside the dev loop. Use when the user says "review my changes", "check this PR", "is this code following our patterns", or invokes /review for a review without the full /develop workflow.

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

8 rules. 0 are common, 6 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 156 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use git to find changed files:

    line 37
  2. Very specific:

    Run the same coverage sweep the in-loop reviewer runs at develop step 06a (fidelity review), so a standalone review is not weaker than the …

    line 45
  3. Distinctive:

    Check each changed file for:

    line 55
  4. Distinctive:

    Blocking: Must fix before merge

    line 65
  5. Distinctive:

    Non-blocking: Should fix, but can defer

    line 66
  6. Distinctive:

    Every changed file reviewed against the five axes in Step 4

    line 86
  7. Very specific:

    Every finding carries a file:line reference

    line 87
  8. Distinctive:

    At completion: Write session summary — mandatory:

    line 101

“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 — Standalone Code Review

Persona delegation

This skill delegates to @code-reviewer. The code-reviewer performs the review (correctness, security, performance, patterns, tests). The skill file provides the review checklist; @code-reviewer provides the depth, failure-mode awareness, and the 15-item false-positive guard.

What this skill does

Reviews code changes for correctness, security, performance, and adherence to project patterns. Use when you need a review without running the full develop workflow.

When to use

  • "Review my recent changes"
  • "Check this PR for issues"
  • "Is this code following our patterns?"
  • Standalone review outside the development cycle

Workflow

Step 1: Load context

@memory-controller load code-reviewer [standalone code review]

Load patterns-and-conventions.md to understand project conventions.

Step 2: Identify changes

Use git to find changed files:

Run git diff --name-only HEAD~1 (or appropriate range)

Step 3: Orphan-AC sweep (02o parity)

Run the same coverage sweep the in-loop reviewer runs at develop step 06a (fidelity review), so a standalone review is not weaker than the in-loop one:

node .agents/scripts/req_to_impl_check.js

Report every acceptance criterion that no implementation and no test references. An unreferenced AC is a defect, not a detail — a criterion nothing implements is a promise the code does not keep. Label the sweep honestly in the report; it is a reference, not a phase-advance gate.

Step 4: Review

Check each changed file for:

  • Correctness: Logic errors, edge cases, null handling
  • Security: Input validation, auth flows, data exposure
  • Performance: N+1 queries, unnecessary allocations, hot paths
  • Patterns: Adherence to project conventions from patterns-and-conventions.md
  • Tests: Coverage of new/changed logic

Step 5: Report

Return findings organized by severity:

  • Blocking: Must fix before merge
  • Non-blocking: Should fix, but can defer
  • Nit: Style/preferences

Include specific file:line references for each finding. Open with one plain sentence stating the overall disposition and the blocking count (DNA 8), then the findings as what's-wrong → why-it-matters → consequence.

Halt conditions

Any one halts the review and escalates to @tech-lead:

  • No changes to review. git diff --name-only over the chosen range returns nothing. A review of an empty diff is a green with nothing behind it.
  • A security finding rated Critical or High. Report it and stop; do not continue the checklist past it as though it were one finding among many.
  • The orphan-AC sweep cannot run. req_to_impl_check.js exits 2 in two cases, and both are fail-closed rather than a pass: artifacts/output/03-strategy/user-stories.md is absent, or it is present but contains no AC-### ids at all — a sweep over zero criteria is not a clean sweep. Report that the sweep did not run; never omit it silently, because "no orphan ACs" and "the sweep never ran" read identically in a report that does not say which happened.

Done when

  • Every changed file reviewed against the five axes in Step 4
  • Every finding carries a file:line reference
  • The orphan-AC sweep has either reported its result or reported that it could not run
  • The report opens with the disposition sentence and the blocking count (DNA 8)
  • @memory-controller session-write has run

Not gated mechanically, and stated rather than implied. This skill produces a review, not an artifact, so there is nothing on disk for a checker to read. Whether the review was thorough is a human judgement — the same limit /code-reviewer carries in the develop loop.

State machine integration

At start: Run node .agents/scripts/orchestrator_state.js status At end: Run node .agents/scripts/orchestrator_state.js complete --agent code-reviewer --artifact code-review

At completion: Write session summary — mandatory:

@memory-controller session-write [agent: @code-reviewer]
Worked on: Standalone code review — {files or feature reviewed}
Decisions: {blocking findings count}, {non-blocking findings count}. Key issues: {1-2 sentence summary}
Next step: {fix blocking issues then re-review, or merge if clear}
Blockers: {any blocking findings or "none"}

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.