resume-craft

# Resume Crafting Skill

Analyze job descriptions and generate tailored, honest, one-page resumes. Produces Markdown source and professional PDF output via bundled script.

## How to Use

**Basic: provide a job description and your background**
```
Tailor my resume for this role:

[paste job description or URL]

My background:
- 5 years as software engineer at TechCorp
- Led team of 4 on cloud migration project
- Python, AWS, Terraform
- B.S. Computer Science
```

**With existing resume**
```
Tailor my resume for this position:

Job: [paste JD]
My resume: [paste or attach current resume]
```

**Career transition**
```
I'm moving from DevOps to product management. Tailor my resume for:
[paste JD]

Transferable experience:
- Led cross-functional infrastructure projects
- Managed stakeholder relationships as TAM
- Built internal tools adopted by 100+ users
```

## Core Principles

1. **Ask questions, don't make things up.** Unsure about a claim? Ask the user. Never fabricate experience, metrics, or titles.
2. **Accuracy over seniority.** A truthful resume beats an inflated one every time.
3. **One page, no exceptions.** Trim bullets, consolidate skills, cut filler.
4. **Match the audience.** A regional bank reads differently than a FAANG. Adjust tone, terminology, and emphasis to fit.
5. **Keywords matter, but naturally.** Incorporate JD language without keyword-stuffing.

## Workflow

### 1. Gather Materials

Check for the user's source resume or experience document. If none exists, ask for:
- Current resume or experience summary
- Target role URL or job description text
- Specific preferences or constraints

**Do not proceed without understanding the user's real background.** If information is vague, ask clarifying questions before writing.

### 2. Analyze the Job Description

- Separate must-have vs nice-to-have requirements
- Note specific technologies, tools, certifications mentioned
- Identify skill gaps between user's experience and JD
- Flag red flags (vague JD, "wear many hats", excessive "fast-paced") and green flags (salary disclosed, clear tech stack, team structure)
- Determine the audience: startup vs enterprise, technical vs business, FAANG vs regional

### 3. Craft the Resume

#### Header
- `# FULL NAME` (18pt bold, centered)
- Contact info on one line, pipe-separated: `City, State | email | phone | linkedin`
- **Never put the target role as a subtitle.** That implies you currently hold that title. Use no subtitle or the candidate's actual title/tagline.

#### Summary (2-3 sentences max)
- Lead with years of experience + core domains + industries
- Keep claims realistic and grounded
- Avoid filler: "proven track record", "results-driven", "passionate", "dynamic"
- Match tone to audience -- "senior technical advisor" not "executive advisor" for non-FAANG

#### Skills (4-5 rows max)
- Lead with skills mentioned in the JD
- Remove skills not relevant to the role
- Format: `**Category:** Skill1, Skill2, Skill3`
- Do NOT use markdown tables -- the PDF parser can't handle them
- Include certifications inline when relevant: `AWS (Solutions Architect Pro)`

#### Experience (reverse chronological)
- **Always list in strict reverse chronological order.** Most recent first. Double-check dates before finalizing.
- Reorder bullets within each role to lead with most JD-relevant
- Rewrite bullets using JD keywords naturally
- Quantify impact: users, %, time saved, $ amounts
- Say "internal users" for internal tools (not "users" which implies external)
- Format: `[Action Verb] + [What] + [How/Result] + [Impact/Metric]`
- Trim older roles to 2 bullets max
- Remove bullets that don't support the target role

#### Certifications & Education
- List relevant certifications by name, not count ("Solutions Architect Pro" not "12 certifications")
- For expired certs: omit dates or say "earned 2020"
- Include degree, school, and year

#### Community / Advocacy (optional)
- One line max. Don't mix unrelated info here.

### 4. One Page Check

If the resume spills to a second page:
1. Trim least relevant bullets first
2. Consolidate skills rows
3. Shorten section headers
4. Remove filler words
5. Reduce spacing only as last resort

### 5. Post-Generation Review

After presenting the tailored resume, provide:

**Strengths:** What makes this candidate competitive for this specific role
**Gaps:** Requirements not fully met and how to address them (courses, projects, reframing)
**Recommendations:** Cover letter hooks, interview talking points, or skills to highlight

### 6. Generate PDF

Convert Markdown to professional PDF using the bundled script:

```bash
pip install reportlab  # one-time setup
python scripts/generate-resume-pdf.py resume.md resume.pdf
```

## Markdown Format for PDF Compatibility

The PDF script parses a specific markdown structure. Follow this exactly:

```markdown
# FULL NAME

City, State | email@example.com | 555-123-4567 | linkedin.com/in/handle

---

## PROFESSIONAL SUMMARY

Two to three sentences. Grounded and specific.

## SKILLS

**Category One:** Skill A, Skill B, Skill C
**Category Two:** Skill D, Skill E, Skill F

## PROFESSIONAL EXPERIENCE

### Company Name | City, State
**Job Title** | Start Date -- End Date

- Quantified achievement bullet
- Another achievement bullet

## CERTIFICATIONS & EDUCATION

**Certifications:** Cert A, Cert B
**Education:** B.S. Field of Study, University Name, Year
```

**Critical rules:**
- `**bold**` for skill categories and role titles, never markdown tables
- Contact info uses ` | ` pipe separators on one line
- `---` only once (after contact info) as header divider
- `## ` for section headers, `### ` for company names, `- ` for bullets

## Special Considerations

**Career changers:** Lead with transferable skills. Reframe past experience using the target role's language. Address the transition in the summary.

**Technical roles:** Include a prominent skills section. List languages, frameworks, tools. Link to GitHub/portfolio if relevant.

**Senior / executive roles:** Focus on strategic impact, team building, revenue. Keep technical details lighter. Can extend to 2 pages if 15+ years.

**Recent graduates:** Lead with education. Include relevant coursework, projects, internships. Emphasize leadership in student orgs.

**Non-traditional backgrounds (no degree, career gap, self-taught):** Lead with experience and certifications. Don't hide gaps -- frame them positively. Certifications and projects speak louder than degrees.

## PDF Spacing Reference

| Element | spaceBefore | spaceAfter |
|---------|-------------|------------|
| Name (H1) | -- | 8pt |
| Contact line | -- | 8pt |
| Section header (H2) | 8pt | 6pt |
| Company header (H3) | 4pt | 1pt |
| Role title | -- | 3pt |
| Body text | -- | 3pt |
| Bullet points | -- | 2pt |
| Skill rows | -- | 2pt |

## Common Mistakes

| Don't | Do |
|-------|-----|
| Put target role as subtitle | No subtitle or actual current title |
| Use markdown tables for skills | `**Category:** items` format |
| "12 certifications" | List by name: Solutions Architect Pro, DevOps Pro |
| "proven track record" | Cut it |
| Claim titles you don't hold | Describe what you did, not what you want to be |
| "C-suite stakeholder management" | "Executive stakeholder engagement" |
| List experience out of order | Reverse chronological -- double-check dates |
| Overclaim AI/emerging tech | "Applied AI", "GenAI prototyping", "AI-driven automation" |
| VP-level tone for all audiences | Match tone to company size and culture |
| Fabricate metrics | Ask the user for real numbers or omit |
| Include every skill you have | Only skills relevant to this JD |
| Generic summary for all jobs | Rewrite summary for each application |

## Tips for Best Results

- **Be specific:** Provide the complete job description, not just a title
- **Share real metrics:** Numbers, percentages, team sizes, dollar amounts
- **Mention constraints:** Page limits, format preferences, sections to include/exclude
- **Iterate:** Ask for revisions, alternative emphasis, or different tone
- **Multiple roles:** Generate separate tailored versions for each application
- **Review before sending:** Always verify accuracy -- the skill optimizes, you verify