job-fact-check
Skill0 starsChanged 18 days agoMITView on GitHub
Truthfulness gate for job-application materials. Use before a resume, cover letter, recruiter message or application-form answer is marked final or handed to the candidate, and whenever a draft adds a new number, title, date or claim.
In this repository4 other files it gives its agents.
- job-form-answersSkill
- job-outreachSkill
- job-pipelineSkill
- resume-builderSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
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
- Very specific: line 8
Every resume, cover letter, outreach note or form answer must trace to the candidate's confirmed facts file (facts_file in …
- Very specific: line 18
Numbers: every number appears in the facts file, or in allowed_numbers with a stated reason (for example "6+ years" computed from dates).
- Distinctive: line 25
Read each bullet or sentence next to the facts file and confirm:
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Job materials fact check
Every resume, cover letter, outreach note or form answer must trace to the candidate's confirmed facts file (facts_file in config/candidate.json). The candidate's latest direct corrections in that file win over older entries. Nothing passes on "probably true".
1. Mechanical check (always run)
python3 skills/job-fact-check/scripts/fact_check.py DRAFT.html [MORE.html|.md|.txt ...] --config config/candidate.jsonIt checks:
- Resumes (file name contains "resume"): each employer's title and dates match
employersin the config exactly. - Numbers: every number appears in the facts file, or in
allowed_numberswith a stated reason (for example "6+ years" computed from dates). - Claims that are commonly inflated: people management, P&L, quota, negotiation, UAT, model training, financial modeling, hype words, plus
extra_banned_claimsfrom the config.
Exit code 1 means stop: fix the wording, rebuild, run again. Never silence a finding by adding an exception unless the candidate confirmed the fact; record the fact in the facts file first.
2. Manual review (the script cannot judge meaning)
Read each bullet or sentence next to the facts file and confirm:
- Ownership – the verb matches what the candidate actually did (proposed, coordinated, built, owned). Work done by colleagues, vendors or account managers is credited to them.
- Attribution – shared results say "contributed to"; approximate results stay approximate ("double-digit" never becomes "23%").
- Scope – a side project is not presented as a job; analysis is not presented as ownership of the downstream operation.
- Consistency – titles, cities, dates and reasons for leaving match every other application; background checks compare them.
- Location and availability match the config.
3. Report
One line per document: PASS or FIX: <what>. Only PASS documents may move to "Materials ready".
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
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.

