resume-analyzer

What this skill does

# Resume Analyzer

Analyze a job description against the master resume (`hojicha/resume.tex`) and candidate context (`hojicha/candidate-context.md`). Output: `hojicha/<company>-<role>-resume/notes.md`.

## Critical Rules

1. **NEVER fabricate experiences, skills, or achievements.** Only use content from the master resume and `candidate-context.md`. You may rephrase and emphasize -- never invent.
2. **Never generate generic content -- ask instead.** If a bullet or gap description would be vague, ask a probing question to get real specifics.

## Workflow

### Step 1: Parse Inputs

Read the master resume, candidate context, and the job description.

```
Required:
- Job description (pasted text, file, or URL -- fetch URL content if needed)
- Master resume: hojicha/resume.tex
- Candidate context: hojicha/candidate-context.md
Optional (ask if not provided):
- Company name
- Role title
- Special instructions (e.g., "emphasize ML experience")
```

Derive `<company>` and `<role>` from the JD. Use lowercase, hyphenated slugs (e.g., `kronos-research-ml-researcher-resume`).

### Step 2: Discovery Interview (Conditional)

Cross-reference JD requirements against candidate materials. If there are 2+ areas where the JD demands depth that current materials don't address:

1. Read `references/candidate-discovery.md` for probing techniques
2. Ask targeted questions one at a time -- wait for each response
3. Maximum two follow-ups per topic
4. Append new discoveries to `hojicha/candidate-context.md`

Skip if `candidate-context.md` already covers the JD requirements. Note in the output if skipped and why.

### Step 3: Keyword Extraction

Extract and categorize keywords from the JD:

| Category | Examples |
|----------|---------|
| Hard skills | Python, PyTorch, distributed training |
| Soft skills | Leadership, cross-functional collaboration |
| Domain knowledge | NLP, reinforcement learning, quantitative finance |
| Tools/platforms | AWS, Docker, Kubernetes |
| Qualifications | BSc in CS, 3+ years experience |

Separate into Required vs Preferred. Count frequency -- terms repeated across the JD are high-priority. Include both acronyms and full forms (e.g., "Natural Language Processing (NLP)").

### Step 4: Gap Analysis

Map each JD requirement to existing resume content:

- **Strong match**: Resume already demonstrates this
- **Reframeable**: Experience exists but needs rephrasing
- **Gap**: No matching experience -- before marking, ask the candidate if they have unlisted experience. Persist any discoveries to `candidate-context.md`

### Step 5: Avenues to Strengthen Application

For each gap, provide:

- **Severity**: High (hard requirement, no evidence), Medium (preferred qualification, missing), Low (nice-to-have, easily bridged)
- **Mitigation strategy**: Specific, actionable recommendation (not generic advice)
- **Why it works**: What signal it sends to the hiring manager

Prioritize High gaps first. Include at least one concrete project idea. Never recommend fabricating experience.

### Step 6: Output

Create `hojicha/<company>-<role>-resume/notes.md` containing:

- JD summary
- Keyword analysis table
- Gap analysis table (requirement, status, resume evidence)
- Avenues to strengthen application
- Recommended section ordering (Education-first for <3yr experience, Experience-first for 3+yr)
- Recommended sections to uncomment/comment from master resume

After completing the analysis, tell the user: "Analysis complete. Run the `resume-tailor` skill to generate the tailored resume.tex from these notes."