Skip to main content

resume-screening

This skill should be used when the user asks to "筛选简历", "筛选匹配岗位", "评估简历与职位匹配", "简历匹配职位", "screen a resume", "screen matching jobs", or "evaluate resume-job fit".

Informations de source

Dépôt
LIghtJUNction/BossKit
Dernière activité de la source
30 juillet 2026 à 03:21
Langue détectée de SKILL.md
anglais
Étoiles
2
Forks
0

Options d'installation

Le prompt qui vérifie d'abord la source est sélectionné par défaut. Vous pouvez passer à une commande directe ou télécharger une copie locale.

Vérifiez les fichiers source

Lisez SKILL.md et les fichiers associés affichés par SkillsMP avant de décider de l'installer.

Affichage de SKILL.md

SKILL.md
Instructions source · Aperçu en lecture seule
name
resume-screening
description
This skill should be used when the user asks to "筛选简历", "筛选匹配岗位", "评估简历与职位匹配", "简历匹配职位", "screen a resume", "screen matching jobs", or "evaluate resume-job fit".
version
0.1.0
# Resume Screening Screen a typed local candidate resume against locally cached jobs through the BossKit command line. Produce reproducible local `manual_review` or `dry_run` plans only. ## Boundaries - Treat the workflow as candidate-side job screening. - Never claim recruiter-side access to candidate pools, employer inboxes, or platform applicant data. - Never greet, chat, apply, submit a resume, or perform any platform resume write. - Never open a browser or use browser automation. - Keep online-resume text ephemeral. Never persist, import, or copy it into a local resume automatically. - Never expose cookies, tokens, platform remote IDs, or PII in evidence. - Describe matching as deterministic title-and-skill screening. Never imply semantic understanding or recruiter-grade ranking. ## Select a Typed Resume 1. List and inspect local typed resumes: ```bash boss resume ls boss resume show <RESUME_NAME> ``` 2. Compare known local revisions when necessary: ```bash boss resume diff <LEFT_NAME> <RIGHT_NAME> ``` 3. Clone a revision only when a separate local screening variant is requested: ```bash boss resume clone <RESUME_NAME> <NEW_NAME> ``` 4. Read the sanitized online snapshot only with: ```bash boss account resume show ``` 5. Request explicit consent before persisting any online-resume-derived content. After consent, require a reviewed strict JSON file and import it explicitly: ```bash boss resume import <PATH> ``` Never treat consent as proof that an automatic online-to-local conversion exists. ## Prepare Cached Jobs 1. Search for relevant AI Agent jobs and cache the results: ```bash boss search "AI Agent" ``` 2. Enumerate and inspect exact cached records: ```bash boss ls boss show <LOCAL_JOB_ID> ``` 3. Deduplicate by company and role. Retain at most one best role per company in the final review set. 4. Exclude known blacklisted companies, job descriptions, and jobs. Inspect configured local blacklist rules when needed: ```bash boss campaign blacklist ls ``` ## Define Hard Gates Create or inspect a reproducible local policy: ```bash boss campaign policy ls boss campaign policy show <POLICY_NAME> boss campaign policy add <POLICY_NAME> --monthly-salary-min 10000 ``` Add only verified include, exclude, welfare, salary, and score rules supported by `boss campaign policy add --help`. Preserve the following decisions in the policy or the manual-review evidence: - Require an AI Agent title or direct role evidence. - Require a monthly salary floor of at least 10K. - Require explicit double-weekend evidence; treat silence or ambiguous schedule wording as unknown. - Record welfare evidence without inferring absent benefits. - Compare experience and education requirements against the selected resume. - Apply location requirements and mismatches explicitly. - Exclude blacklist matches. Classify every unknown hard requirement as `manual_review`, never as pass. Classify a verified hard-gate violation as excluded. Keep preferences separate from hard gates. ## Screen and Review 1. Run deterministic local screening: ```bash boss campaign screen --resume <RESUME_NAME> --policy <POLICY_NAME> ``` 2. Treat the generated ranking as a title-and-skill signal only. Inspect each retained cached job with `boss show <LOCAL_JOB_ID>` before accepting its evidence. 3. Review generated local plans: ```bash boss campaign plan ls ``` 4. Summarize local workflow counts only after screening: ```bash boss campaign stats ``` 5. Leave all results in `manual_review` or `dry_run`. Do not transition the workflow into greeting, chatting, application, or resume submission. ## Report Reproducible Evidence Return a compact screening record: ```text mode: candidate_side_local_screening policy: <name and effective rules> resume: <local name and revision identity> cached_jobs: <local cached job IDs inspected> scores: <deterministic scores and matched title/skills> excluded: <job, gate, and supporting evidence> unknowns: <job and unresolved hard requirements> manual_review: <deduplicated retained jobs> dedupe: <company/role duplicate decisions> counts: <screened, excluded, unknown, retained> ``` Keep salary, schedule, welfare, experience, education, location, and blacklist evidence traceable to the selected resume, policy, or exact cached job. Mark missing evidence as unknown. Never report the deterministic matcher as proof of overall suitability.
Voir sur GitHub