agentleFS
leoleolroleoleolro/career-os/.claude/skills/hunt/SKILL.md

hunt

Skill0 starsChanged 56 days agoMITView on GitHub

Run the job hunt across all sources, rate real fit for every new posting, and shortlist. Use when the candidate says "hunt", "find jobs", "any new jobs", "what's out there", "/hunt", or "hunt review".

15 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

3 rules. 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 11 edited versions we read keep these.

  • 1

    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:

    If a source logs 429/blocked, note it and continue — LinkedIn and Seek can be searched manually with the Browser pane …

    line 15
  2. Very specific:

    Update data/jobs.json: for each rated job set rating ("Strong"|"Good"|"Stretch"|"Skip") and fit (one sentence: the reason).

    line 36
  3. Distinctive:

    Never fabricate a posting.

    line 49

“Common” is measured: kept by at least 6 in 10 of the 11 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

/hunt

1. Run the fetch (deterministic)

export PATH="$HOME/.local/lib/node/bin:$PATH"
cd "$CAREER_OS"   # your career-os folder
node hunt.mjs --new-only --details 15 --min-score 40 2>&1 | tail -60

Takes ~3–5 minutes (sources run in parallel; each is paced per host). If tools/daily.sh already ran today (see data/logs/), skip straight to reading the report. If the candidate said hunt review, skip the run and just read today's report. If a source logs 429/blocked, note it and continue — LinkedIn and Seek can be searched manually with the Browser pane (https://www.linkedin.com/jobs/search/?keywords=…&location=Perth / https://www.seek.com.au/…-jobs/in-Perth-WA) when the adapter is rate-limited.

1b. Browser-only sources (2 minutes)

McKinsey renders only in a real browser (Akamai). Open https://www.mckinsey.com/careers/search-jobs?countries=Australia in the Browser pane, get_page_text, and compare the Consulting titles against data/jobs.json (currently tracked: Business Analyst 15136, Experienced Junior Associate 101796, Associate 15264). Record any new ANZ consulting role with a small node script using normalizeJob + mergeJobs (see sources/mckinsey.mjs header for field shapes). APSJobs (federal government, Salesforce app): open https://www.apsjobs.gov.au/s/job-search?keyword=<query> in the Browser pane and record relevant roles the same way; JobsWA (state) is covered by the wagov adapter. Bain and Indeed are the other browser-only checks when the candidate asks.

2. Read the report

reports/hunt-<today>.md — the "New since last run" section. Also read profile/preferences.json (gates) and profile/master-resume.md (what the candidate can honestly claim).

3. Rate the jobs that matter (you, not the prescore)

Scope — do NOT try to rate hundreds: (a) every job in the report's "New AI / Oracle / consulting roles in your country" section, (b) the top 20 of "Other new postings", (c) any dream-company job. First pull their full descriptions: node tools/details.mjs --top 40 (fetches for the highest-prescore jobs lacking one; Seek/LinkedIn/BCG/ORC/boards all supported). Then assign:

  • Strong — the candidate meets the stated requirements; role advances a target path (AI engineer / FDE / AI consultant, Oracle EPM, consulting, data). Apply this week.
  • Good — apply. One bridgeable gap, including a "3–5+ years" line: years asked are a preference, not a wall (the candidate's rule, 2026-09-15; 66 of 121 AU AI ads don't even state one).
  • Stretch — apply if time allows, and network: 5–7 years asked, or 2+ real gaps, or a dream company above the candidate's level.
  • Skip — 8+ years, principal/staff/manager titles, wrong domain, or a true hard gate (citizenship/clearance/degree field). Human factor: rate what the candidate could plausibly do, not what the ad's HR template says. Mining-sector, AI-delivery and consulting evidence can outweigh a years line.

Priority is separate from fit. priority (P1/P2/P3, computed by lib/score.mjs upside() from preferences.json → priority_model) answers "is it worth the candidate's effort": brand direction vs the current employer (MBB/AI-native = up; Big 4 peers = lateral; Perth MSPs/boutiques = down), interest (AI/FDE/agentic, MBB consulting = high; generic BA/analyst = low), pay vs current package, contract penalty. the candidate already has a stable job — a same-pay, less prestigious, less interesting role is P3 even if it is a Strong fit. When you shortlist, order by priority then rating, and say in one line why a P2/P3 Strong is still (or isn't) worth applying to. If the tiers misclassify a company, edit company_tiers rather than overriding by hand.

Hard gates: graduate programs → work rights (the candidate is an Australian citizen — OK) and "graduated within N years" (Dec 2023). Location: Perth preferred; Sydney/Melbourne/Brisbane/Canberra/Adelaide acceptable (the candidate will relocate for the right role); overseas only for dream companies (Anthropic, OpenAI) and only if the posting doesn't require local work rights. Never rate a job Strong on the title alone — read the description (--details fetched it into data/jobs.json).

4. Write ratings back

Update data/jobs.json: for each rated job set rating ("Strong"|"Good"|"Stretch"|"Skip") and fit (one sentence: the reason). Use a short python3/node snippet; keep all other fields.

4b. Refresh derived views

node tools/rescore.mjs && node tools/insights.mjs && node tools/shortlist.mjs && node tools/dashboard.mjs — insights (reports/market-insights-<today>.md) shows what the market asks for vs the candidate's resume; mention any new gap that crosses 20% of AU AI ads.

5. Report to the candidate (compact)

  • Sources health (fetched/errors per source, from the report header).
  • Top 5 shortlist ordered by priority: P1 title — company — location — link — rating — one-line why — recommended next action (/tailor <url>, "network first", "wait for 2027 intake").
  • Anything time-sensitive (closing dates, "posted today").
  • Then ask which to /tailor. Do not tailor, apply, or contact anyone without being asked.

Hard stops

  • WebFetch of ATS pages (Phenom/BCG, Workday, Oracle ORC) often returns a "position has been filled" decoy. Verify with the Browser pane or the adapter's details() before telling the candidate a role is closed.
  • Never fabricate a posting. Every link you give must come from the report or a page you opened.

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.