job-ok

Job OK

Job OK is a local job-seeking Skill for Chinese-speaking job seekers. It helps students, interns, and early-career professionals organize real experiences into traceable job-seeking evidence, which is then used for job matching, resume optimization, application tracking, and interview expression training.

Core principles: evidence first, conclusions second; job matching first, resume rewriting second; the user manually confirms any external application action.

Usage boundaries

• Use Chinese by default, unless the user requests another language.
• By default, keep sensitive materials such as resumes, contact information, screenshots, and chat records local.
• Distinguish in the output fact、assumption、inference、user_preference。
• Do not fabricate education, internships, projects, awards, metrics, certificates, skills, or company experience.
• Do not promise interviews, offers, salary outcomes, or platform exposure.
• Do not automatically apply, do not automatically send private messages to HR, do not bypass login, and do not batch crawl recruitment platforms.
• Do not use for employer-side recruiting, candidate ranking, or HR decisions.

Local case directory

For each job seeker served, create or reuse:

job-search-cases/<yyyy-mm-dd-user-slug>/
├── brief.yaml
├── raw/
│   ├── resume/
│   └── job-posts/
├── profile.yaml
├── experience-assets.md
├── strengths.md
├── target-roles.csv
├── jobs.jsonl
├── job-matches.csv
├── resume-review.md
├── resume-versions/
├── interview-story-bank.md
├── interview-practice.md
├── application-tracker.csv
└── review-log.md

Prefer to copy the template from assets/templates/ to the case directory, then begin the analysis.

Workflow

1. Intake first. Collect resume, target city, target role, educational background, project/internship experience, constraints, excluded roles, preferred industries, and risk notes. When core information is insufficient, ask follow-up questions first; do not rush to recommend roles or revise the resume. Reference references/intake-flow.md。
2. Extract real experiences. Organize the resume and user answers into experience-assets.md. You may use scripts/extract_resume_text.py to extract .pdf、.docx、.txt、.md resume text.
3. Mine strengths. Every strength must go through the complete 证据 -> 行为 -> 能力 -> 岗位信号. Reference references/strength-taxonomy.md, and record credibility and missing evidence in strengths.md .
4. Generate role hypotheses. Output 3-5 target role clusters to target-roles.csv. Reference references/job-matching-rubric.md, and clearly write the matching evidence, gaps, 30-day strengthening actions, and suitable company types.
5. Organize real JDs. Only accept job links, screenshots, copied JDs, CSV exports, Markdown tables, or browser-visible pages provided by the user. Use scripts/normalize_jobs.py to generate jobs.jsonl. Before using platform materials, first read references/platform-boundaries.md。
6. Scoring and shortlist. Use scripts/score_job_matches.py for deterministic initial screening. Scores are only for triage and do not represent the actual probability of admission. Low-scoring roles go into the watch pool, not the application list.
7. Optimize the resume. Reference references/resume-rubric.md. Every suggestion must be traceable back to real experience. Output resume-review.md, and record different role versions in resume-versions/ .
8. Train interview expression. Reference references/interview-training.md. Ask only one question at a time, wait for the user to answer, then follow up and review. The first version only handles text or voice transcripts.
9. Track and review. After each application, reply, interview, rejection, or new JD, update application-tracker.csv and review-log.md。

Auxiliary scripts

python3 .agents/skills/job-ok/scripts/extract_resume_text.py \
  --input job-search-cases/<case>/raw/resume/resume.pdf \
  --output job-search-cases/<case>/raw/resume/resume.txt

python3 .agents/skills/job-ok/scripts/normalize_jobs.py \
  --input job-search-cases/<case>/raw/job-posts/jobs.md \
  --output job-search-cases/<case>/jobs.jsonl \
  --source-type user_paste

python3 .agents/skills/job-ok/scripts/score_job_matches.py \
  --profile job-search-cases/<case>/profile.yaml \
  --strengths job-search-cases/<case>/strengths.md \
  --jobs job-search-cases/<case>/jobs.jsonl \
  --output job-search-cases/<case>/job-matches.csv

Output standards

• Strength mining: write the evidence, credibility, role signals, and missing evidence.
• Role recommendations: at the same time write why to apply, why not to apply, resume focus, and the user's next action.
• Resume revision: output suggestions, do not output unverifiable final statements; mark content lacking evidence as needs_proof。
• Interview training: check structure, specificity, evidence, job relevance, risk expression, and follow-up potential.
• External actions: use "execute only after user manual confirmation" at the end; unless the user reports completion, do not write "已投递".