- name
- resume-screener
- description
- Screen resumes against a job description. Scores candidates on qualification alignment, experience match, and fit. Use when HR uploads resumes for a job opening.
- license
- Apache-2.0
- metadata
- {"author":"Red Hat ET","category":"human-resources"}
# Resume screener skill
When asked to screen or grade resumes against a job description:
1. Use `read_file` to read the job description document first
2. Extract the key requirements:
- **Required qualifications** (must-have skills, certifications)
- **Preferred experience** (years, domains, technologies)
- **Cultural fit indicators** (leadership style, collaboration)
3. For each resume:
- Use `read_file` to read the resume
- Score on a 1-10 scale for each category:
- **Qualification alignment**: how well required skills match
- **Experience match**: relevance and depth of experience
- **Fit**: alignment with role expectations and culture
- Compute a weighted overall score (qualifications 40%,
experience 35%, fit 25%)
- Write a one-paragraph justification
4. Produce the final output as a ranked table:
| Rank | Candidate | Overall | Quals | Exp | Fit | Justification |
|------|-----------|---------|-------|-----|-----|---------------|
5. If there are more than 20 candidates, also provide a short list
of the top 20 with a recommendation paragraph for each
6. Use `write_file` to save the results if requested
## Important guidelines
- Never disqualify candidates based on protected characteristics
- Flag gaps or concerns rather than making binary accept/reject
decisions — the recruiter makes the final call
- If a resume is in a format you cannot parse, note it as
"unreadable" with the filename and move to the next
- When qualifications are ambiguous (e.g., "5+ years" vs
"4 years 11 months"), give the candidate the benefit of the doubt
View on GitHub