agentleFS
sree-datasree-data/job-search-agent/.claude/skills/find-jobs/SKILL.md

find-jobs

Skill0 starsChanged 2 months agoMITView on GitHub

Pull new data engineering postings from Dice, ZipRecruiter, and LinkedIn job-alert emails, score them against targets/criteria.md, and add the good ones to pipeline/jobs.csv

11 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use apply-to-job and interview-prep.

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.
  • mukewlSkill

    apply-to-job

    End-to-end prep for one job application - read the posting, produce a tailored page-checked CV in the user's own formatting, find 5-10 named people worth contacting at that company, and draft LinkedIn messages for the user to send by hand. Use when given a job URL or job description and asked to apply, prepare an application, tailor a CV for a role, or find who to reach out to about a job.

    Also in 3 projects that use this

  • AnthropicSkill

    interview-prep

    Create structured interview plans with competency-based questions and scorecards. Trigger with "interview plan for", "interview questions for", "how should we interview", "scorecard for", or when the user is preparing to interview candidates.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

8 rules. 0 are common, 5 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 11 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

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

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

    Read targets/criteria.md, targets/companies.md, and targets/market-demand.md.

    line 8
  2. Very specific:

    Search Dice and ZipRecruiter for each title in criteria, nationwide and remote.

    line 9
  3. Distinctive:

    Run all three queries:

    line 14
  4. Very specific:

    All three are needed because forwarding rewrites the sender inconsistently.

    line 19
  5. Distinctive:

    Keep the content terms bound to query 2 so it matches alerts rather than everything that account has ever sent.

    line 22
  6. Distinctive:

    If these queries return nothing, say so explicitly in the step 6 output rather than staying silent, and name which of the three matched …

    line 31
  7. Distinctive:

    If a connector is unavailable this session, run with the sources that are up rather than stopping, and say which source was missing in the …

    line 42
  8. Distinctive:

    Don't apply to anything. Don't pad the list.

    line 45

“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

Find jobs

  1. Read targets/criteria.md, targets/companies.md, and targets/market-demand.md. Read pipeline/jobs.csv so you don't add duplicates (match on company + title, or URL).
  2. Search Dice and ZipRecruiter for each title in criteria, nationwide and remote. Limit to postings from the last posted_within_days. Dice pages carry the full description in structured data; fetch them. ZipRecruiter results have no description text, so open the posting URL when one exists. 2a. Search Gmail for LinkedIn job alerts from the last 3 days, both direct and forwarded. LinkedIn sends Alex's alerts to alex.rivera.alerts@example.com, a separate account that he forwards into the connected inbox (alex.rivera@example.com). Note the two addresses differ only in the order of the name halves; they are different accounts, not a typo. Run all three queries:

    • from:linkedin.com newer_than:3d
    • from:alex.rivera.alerts@example.com (linkedin OR "job alert" OR "jobs for you" OR "new jobs") newer_than:3d
    • (subject:Fwd OR subject:Fw OR "Forwarded message") ("job alert" OR "jobs for you" OR "new jobs" OR linkedin) newer_than:3d

    All three are needed because forwarding rewrites the sender inconsistently. Gmail auto-forwarding usually preserves LinkedIn as the From, so query 1 catches it; a manual forward puts the forwarding account in From, so query 2 catches it; query 3 is the fallback for any other shape. Keep the content terms bound to query 2 so it matches alerts rather than everything that account has ever sent.

    Open each hit with get_thread at PLAIN_TEXT and pull every job URL out of the body, including from a quoted forward block. LinkedIn alert links appear as linkedin.com/jobs/view/<id> and as tracking links under linkedin.com/comm/jobs/view/<id>; normalise both to linkedin.com/jobs/view/<id>. Fetch each public posting page (allowed without logging in, per hard rule 1) and treat those postings exactly like job-board results.

    If these queries return nothing, say so explicitly in the step 6 output rather than staying silent, and name which of the three matched zero. A sender address that has never sent mail looks identical to a quiet week, and only the first is worth fixing.

  3. For each posting, read the full description and check:
    • level, per criteria
    • sponsorship signal: yes if it says sponsorship available or the company has H-1B filing history; no if it matches any skip_posting_if phrase; unknown otherwise. Skip no.
    • stack overlap with profile/background.md
  4. Score 0 to 100 using the weights in criteria, then run JobFitRanker from the engine and use its score as the tiebreaker. Only add postings at or above minimum_score_to_pursue.
  5. Append each new posting to pipeline/jobs.csv with the next id, today's date, status new, and a one-line note on why it scored what it did.
  6. Print a short table: id, company, title, location, score, sponsorship signal. Sorted by score. Recommend the top 3 to pursue today and say why in one line each.

If a connector is unavailable this session, run with the sources that are up rather than stopping, and say which source was missing in the step 6 output so the day's list can be read for what it is.

Don't apply to anything. Don't pad the list. Five good roles beat thirty mediocre ones.

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.