agentleFS
HimadriTryingHimadriTrying/ai-job-search-agent/.claude/skills/job-scout/SKILL.md

job-scout

Skill1 starsChanged 3 months agoMITView on GitHub

Discover and score open roles against the user's rubric. Sweeps ATS APIs, drops out-of-band roles before scoring, sorts survivors into apply-first / worth-a-look / skipped. Runs scheduled (daily digest) or on demand.

6 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

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

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 0

    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. Very specific:

    Write a dated digest to data/digests/YYYY-MM-DD.md and (if scheduled) commit it.

    line 26
  2. Very specific:

    Run it: python run.py (live) or python run.py --offline jobs.json (no network).

    line 49

“Common” is measured: kept by at least 6 in 10 of the 4 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

job-scout — discovery + scoring

Precedent

Sweeps public ATS APIs (Greenhouse, Lever, Ashby, SmartRecruiters), hard-drops out-of-band roles before any LLM call, then scores survivors with a cheap model. Cost scales with matches, not listings.

Data sources (public ATS APIs — no scraping)

Greenhouse, Lever, Ashby, SmartRecruiters each expose public job-listing endpoints. Maintain a watchlist of target companies (see data/ — create watchlist.txt). A full sweep of ~100 companies takes about a minute.

Pipeline

  1. Sweep the watchlist via ATS APIs → raw listings.
  2. Keyword pre-filter BEFORE any LLM call — cheap. Drop obviously off-target roles.
  3. Hard drops from profile/04-job-evaluation.md: min_seniority (drop below Senior), hard-cued experience minimums, location, work-auth, excluded industries.
  4. Score survivors against the rubric (mostly penalties). Use a cheap model for scoring, a stronger one only for anything you tailor later — deliberate cost tiering.
  5. Sort into apply first / worth a look / skipped, one-line reason each.
  6. Write a dated digest to data/digests/YYYY-MM-DD.md and (if scheduled) commit it.

Reliability

Malformed scoring output retries once, then marks the role unscored rather than crashing the whole run.

Mode

Scheduled (daily, cheap, benefits from overnight) via .github/workflows/daily-scout.yml, or on demand when the user asks "what's out there".

Implementation (built and tested)

Real code lives in scout/ and run.py:

  • scout/ats.py — keyless public-JSON clients for Greenhouse/Lever/Ashby/SmartRecruiters, normalized to one schema. Companies are ats:token (e.g. greenhouse:stripe).
  • scout/seniority.py — title→level ladder with the inverted floor (drops below Senior).
  • scout/filters.py — keyword prefilter + hard drops (experience gate fires only on hard-cued minimums; location; excluded industries).
  • scout/scoring.py — penalty-first scorer → score, reasons, bucket.
  • run.py — orchestrates the pipeline and writes data/digests/YYYY-MM-DD.md.

Config: copy config.example.yaml → config.yaml, set your real years/seniority/locations. Watchlist: copy watchlist.example.txt → watchlist.txt, one ats:token per line.

Run it: python run.py (live) or python run.py --offline jobs.json (no network). Tests: python tests/test_scout.py — all green. To run live you need network access from wherever this executes (works locally and in the GitHub Action; not in a sandbox with egress disabled). Endpoints are current public feeds but slugs drift — a 404 means update the token.

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.