agentleFS
oleg-chibikovoleg-chibikov/agent-skills/skills/review/SKILL.md

review

Skill1 starsChanged 2 months agoMITView on GitHub

Use only when the user explicitly asks to review a pull request: "review this PR", "code review", "review" plus a PR link or PR number, or the same in another language.

192 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use close-spec, design-with-images 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

  • 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

    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

  • 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

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

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 157 edited versions we read keep these.

  • 3

    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:

    Every finding MUST be usable by someone outside this repository.

    line 8
  2. Distinctive:

    Every step names the file holding its detail.

    line 11
  3. Very specific:

    Read LANGUAGE.md next to this file.

    line 16
  4. Distinctive:

    All three MUST be read before the first line of the review.

    line 24

“Common” is measured: kept by at least 6 in 10 of the 157 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 code

Every finding MUST be usable by someone outside this repository. Jargon, internals and "consider refactoring" MUST NOT appear.

Every step names the file holding its detail. You MUST read that file when you reach the step, before writing a line of what the step produces.

1. Set the language and the writing rules

  • Read LANGUAGE.md next to this file. The language it names is the report language, and it MUST beat the language of the user's ask. No such file? Write English.
  • Which part comes back in which language: references/report-language.md.
  • How every line is written: references/writing-rules.md.

All three MUST be read before the first line of the review.

2. Find what to review

The file, PR or branch the user named, else uncommitted changes, else the branch against its base. You MUST NOT ask when the target is obvious. The order, the commands and when you MAY ask are in references/find-the-target.md.

3. Read the repo rules first

Repo rules beat your habits, and what the repo requires MUST NOT be flagged. The files to open are in references/repo-rules.md.

4. Trace the flow, map it as you go

What a person does to make a suspicious line run, and how the changed files hang together. How far to chase it is in references/trace-the-flow.md.

5. Look for what matters

Six things in order, breakage first and style last, plus the duplicate check that runs on every review: references/what-to-look-for.md.

6. Write the report

Four parts in this order, each with its own file:

  1. What this change is about: references/part-1-summary.md.
  2. The map and the reading order: references/part-2-map.md.
  3. The findings table: references/part-3-table.md.
  4. The findings, first every comment with where it goes, then the explanation of each: references/part-4-findings.md, plus references/finding-shape.md for the blocks each one is built from.

The rules holding over all four, and what MUST NOT be added to them, are in references/report-shape.md. Read it before part 1.

7. Run the second round: architectural gaps

The four parts go on screen first. Then one more pass over the same change, asking whether the design fits the task at all and what it leaves out, before anything is posted: references/architecture-pass.md.

8. Offer to post the comments

The second round MUST close with one offer: put the comments on the PR, the findings of both rounds together. Nothing goes on the PR until the user asks for it. The offer, the question that comes before the first call and the gh call per comment are in references/posting-comments.md.

9. Offer the deep dive

Once the posting is done or the user closed it, offer which findings to dig into. This is the last question of the review. The choices, what digging into one means and how the finding comes back are in references/deep-dive.md.

10. Before you send

You MUST read references/final-checklist.md and run it over the whole answer. Nothing goes out before that.

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.