agentleFS
priyansh1210priyansh1210/shiro/.claude/skills/review/SKILL.md

review

Skill1 starsChanged 6 days agoMITView on GitHub

Run the rejection panel on a resume for one job. Picks the best base resume to start from, ten reviewers (eleven for design roles) try to reject it in parallel, a debrief merges their feedback into a fix plan, and the resume writer produces a tailored, truthful new version.

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
  • Other versions192 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 192

    1. garrytan/gstack★ 142ksame name
    2. ansible/ansible★ 71ksame name
    3. docling-project/docling★ 71ksame name
    4. traefik/traefik★ 66ksame name
    5. ClickHouse/ClickHouse★ 51ksame name
    6. temporalio/temporal★ 24ksame name
    7. ccch1mneyyy/dsh-Go★ 24ksame name
    8. tisfeng/Easydict★ 15ksame name
    9. juicedata/juicefs★ 15ksame name
    10. elie222/inbox-zero★ 13ksame name

    182 more not listed

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository4 other files it gives its agents.

The file

Show its rules and text

We split this skill into the rules it gives an agent

13 rules. 0 are common, 6 are distinctive, and 7 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.

  • 7

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Save tokens: never read a resume PDF if a text copy exists at memory/resume/.text/<file>.txt.

    line 14
  2. Distinctive:

    File path: read it.

    line 17
  3. Distinctive:

    Nothing given: ask the user to paste the JD and wait.

    line 19
  4. Very specific:

    Write jd.md: the full job description text with page clutter removed (navigation, cookie banners, footers).

    line 26
  5. Distinctive:

    A resume path was given in the arguments: use it. Skip the rest of this step.

    line 29
  6. Distinctive:

    None: ask the user for a resume path.

    line 31
  7. Very specific:

    Several: run the resume-picker agent and wait for it:

    line 33
  8. Distinctive:

    Use the resume it picked, and tell the user in one line which one and why.

    line 38
  9. Very specific:

    If memory/ doesn't exist, tell the user that /setup gives much better results (the writer can use their whole history, not just one …

    line 40
  10. Very specific:

    Write resume-original.md: a faithful text copy of the chosen resume, with the same sections, wording and order.

    line 43
  11. Distinctive:

    Wait until every reviewer has reported back before going on: If an agent fails, retry it once.

    line 59
  12. Very specific:

    Run the debrief agent and wait for it:

    line 64
  13. Very specific:

    Run these two commands, each on its own, with nothing else on the line (no cd):

    line 82

“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: the rejection panel

Arguments: $ARGUMENTS

Each agent's effort comes from its definition (effort: in .claude/agents/*.md): Low for quick reviewers, Medium where judgment needs depth, High for the writer. Don't override it unless the user asks. (The Shiro app doesn't use this skill: it runs its own cheaper pipeline, see ui/review-pipeline.mjs.)

Save tokens: never read a resume PDF if a text copy exists at memory/resume/.text/<file>.txt; otherwise run node tools/resume-text.mjs <file> once (it caches the text). Agents only ever read text files.

1. Get the job description

  • File path: read it.
  • URL: fetch it with WebFetch. Career pages are often rendered by JavaScript, so if the text comes back empty or without the actual job, and the Claude in Chrome tools are available, open it in a new tab and use get_page_text. Otherwise ask the user to paste the JD.
  • Nothing given: ask the user to paste the JD and wait.

Extract the company name and the role title. If the company isn't clear, ask.

2. Create the run folder

runs/<YYYY-MM-DD>-<company>-<role>/: lowercase, hyphens, short. If it already exists, add -2, -3, and so on.

Write jd.md: the full job description text with page clutter removed (navigation, cookie banners, footers). Change nothing else. First line: Source: <URL or path>.

3. Pick the resume to start from

  • A resume path was given in the arguments: use it. Skip the rest of this step.
  • Otherwise, collect the base resumes: every file listed under ## Resumes in memory/profile.md, plus any other resume file in memory/resume/.
    • None: ask the user for a resume path.
    • One: use it.
    • Several: run the resume-picker agent and wait for it:

      Job description: <run>/jd.md Resumes: <path 1>, <path 2>, … Write your pick to: <run>/pick.md

      Use the resume it picked, and tell the user in one line which one and why.

If memory/ doesn't exist, tell the user that /setup gives much better results (the writer can use their whole history, not just one resume), and continue if they want to.

4. Copy the resume as text

Write resume-original.md: a faithful text copy of the chosen resume, with the same sections, wording and order. Don't fix or improve anything. First line: Source file: <path>. Use the cached text (memory/resume/.text/<file>.txt, or run node tools/resume-text.mjs <file>) rather than reading the PDF.

5. Run the panel: all at once

Send one message containing one Agent call per reviewer so they run in parallel and none of them sees another's opinion. Subagent types:

ats, recruiter, hiring-manager, teammate, rival-applicant, ceo, technical-interviewer, skeptic, skip-level-manager, hr-screener

Add design-lead only when the job is a design role (product, UX, UI, interaction, visual or service design, design research). It judges design craft, so on other jobs it only adds noise and cost. Tell the user in one line whether it's in the panel.

Prompt for each (fill in the paths; don't paste the resume or JD text into the prompt, the agents read the files):

Review this resume for <Role> at <Company>. Job description: <run>/jd.md Resume text: <run>/resume-original.md Write your report to: <run>/feedback/<agent-name>.md

Wait until every reviewer has reported back before going on. If an agent fails, retry it once. If it fails again, continue without it and tell the user.

Show the user the scoreboard built from the one-line replies: a table of reviewer, verdict, score, reason.

6. Debrief

Run the debrief agent and wait for it:

Feedback folder: <run>/feedback/ Job description: <run>/jd.md Resume text: <run>/resume-original.md Write the debrief to: <run>/debrief.md

7. Rewrite

Run the resume-writer agent and wait for it:

Run folder: <run>/ Debrief: <run>/debrief.md Job description: <run>/jd.md Resume text: <run>/resume-original.md Resume pick notes (if any): <run>/pick.md Career memory: memory/

8. Make the HTML and PDF

Run these two commands, each on its own, with nothing else on the line (no cd):

node tools/resume-html.mjs <run>/resume.md <run>/resume.html node tools/html-to-pdf.mjs <run>/resume.html <run>/resume.pdf

They turn the writer's resume.md into the printable page and the PDF, using the Chrome or Edge on this computer. Both cost nothing. If it fails, say so in the report; the user can open resume.html and press Ctrl+P → Save as PDF instead.

9. Report to the user

Keep it short:

  • Which resume it started from (and why, if the picker chose).
  • Average score and the ADVANCE / MAYBE / REJECT counts.
  • The 3–5 most important fixes that were made.
  • Full paths to resume.pdf, resume.md, changes.md, questions.md and debrief.md. /apply uploads resume.pdf from here for this job.
  • The questions from questions.md, and that they can answer them with /remember.
  • Offer round 2: run the same reviewers on the new resume.md to see whether the score went up. If they say yes, repeat step 5 with <run>/resume.md as the resume text (and no original file), writing to <run>/round-2/feedback/, then show the before/after scoreboard side by side. Don't rewrite again unless they ask.
  • Offer to save it as a base resume for this kind of role: copy resume.pdf into memory/resume/<role>.pdf (or, if you can't run shell commands, use the run's path) and add a line - <Role>: <path> under ## Resumes in memory/profile.md, with <Role> kept to a few words (e.g. AI Engineer Intern (Acme)), never a list of what changed. Future jobs of this kind will start from 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.