Hr Resume Evaluation

Use when a recruiter wants to batch-evaluate resumes, run general resume quality checks, optionally match resumes to a JD and company background, and produce evidence-backed Markdown and CSV review reports.

Core Rule

Use this Skill only as a recruiter review aid. Never present scores as final hiring, rejection, elimination, or admission decisions.

Before Use

1. Put resumes in a local input folder. Supported resume formats: Markdown, TXT, PDF, DOCX.
2. Optionally prepare a JD file and company background file in Markdown or TXT.
3. Set the API key environment variable configured in config/model.yaml, default DEEPSEEK_API_KEY.
4. Read references/scoring-rubric.md, references/role-family-taxonomy.md, references/compliance-policy.md, and references/output-schema.md.

Command

python scripts/evaluate_resumes.py `
  --resumes input/resumes `
  --jd input/jd.md `
  --company input/company.md `
  --output reports `
  --config config/evaluation.yaml `
  --model-config config/model.yaml

Omit --jd and --company for general resume quality review only.

Required Outputs

• One Markdown report per successfully evaluated resume.
• One batch summary Markdown report.
• One CSV summary file.
• Clear failure records for unreadable or invalid files.

Safety Rules

• Do not score protected or job-irrelevant traits.
• Do not infer missing resume facts.
• Do not perform background investigation.
• Do not automatically reject or approve candidates.
• Treat model output as advisory until reviewed by a human recruiter.