Resume Tailor

Resume Tailor Skill

Orchestrates a sequential pipeline of specialized agents to analyze a resume against a job description, then produces a tailored resume with gap analysis, ATS scoring, change tracking, and actionable next steps.

Commands

• /resume tailor — Full pipeline: analysis, gap report, rewrite, diff, ATS score, synthesis
• /resume gap — Gap analysis only (no rewrite)
• /resume ats — ATS compatibility score only
• /resume cover — Generate cover letter (3 variants: standard, short, talking points)
• /resume interview — Interview prep from gap analysis + JD
• /resume compare — Compare resume against multiple JDs with fit matrix

Input Requirements

Before starting, collect:

1. Resume — paste as text, or provide a .pdf/.docx file path
2. Job Description — paste full JD text or provide a file path
3. Target role title (if not clear from JD)
4. Optional: Company name, submission channel (ATS portal / email / referral), specific concerns ("I lack X skill")

If any required inputs are missing, ask for them before proceeding. Do NOT proceed with partial inputs.

For /resume compare, collect 2-5 job descriptions instead of one.

Full Tailor Flow (/resume tailor)

Step 1: Structured Analysis (Replaces Python Parser)

Analyze the resume and JD directly in-context. Do NOT use any external scripts. Extract the following structured data by reading both documents carefully:

From the Resume:

• All technical skills (languages, frameworks, tools, platforms, databases, methodologies) — extract everything mentioned, not just from a predefined list
• All soft skills and leadership indicators (team size, mentoring, cross-functional work)
• Experience timeline: for each role, extract company, title, start date, end date. Calculate cumulative experience handling overlapping roles correctly (overlapping months count once, not twice)
• Section-by-section breakdown: summary, experience, skills, education, projects, certifications
• Quantified achievements: any numbers, percentages, scale indicators

From the Job Description:

• All requirements classified as:
  • Core Technical — Primary tech stack (weighted highest)
  • Supporting Technical — Infrastructure, tooling, adjacent tech
  • Process & Methodology — Agile, code review, testing practices
  • Soft Skills — Communication, leadership, collaboration
  • Nice-to-Have — Explicitly optional items
  • Boilerplate — Generic filler ("team player", "fast-paced environment") — ignore these
• Years of experience required
• Seniority signals (IC vs lead vs manager)
• Domain/industry requirements

Keyword Overlap (with semantic matching):

• Exact matches between resume and JD
• Terminology mismatches (same skill, different name — e.g., "JS" vs "JavaScript")
• Truly missing skills (not in resume at all)
• Semantic matches (e.g., resume says "led a team of 5" → JD says "leadership experience")

Store this analysis in memory for use by all downstream agents.

Step 2: Gap Analyzer Agent

Read agents/gap-analyzer.md and run it with:

• Full resume text
• Full JD text
• The structured analysis from Step 1

This agent produces:

• Candidate profile (career stage, move type)
• JD requirement classification (what's truly required vs aspirational)
• Hard gaps with severity ratings (DEAL-BREAKER / SIGNIFICANT / MINOR)
• Soft gaps with rewrite suggestions
• Keyword gaps (true gaps vs terminology mismatches)
• Strengths to emphasize
• Overall match assessment

The gap analysis output feeds all downstream agents. This step must complete before proceeding.

Step 3: Bullet Rewriter Agent

Read agents/bullet-rewriter.md and run it with:

• Full resume text
• Full JD text
• Gap analysis output from Step 2

This agent produces:

• Rewritten summary (if applicable)
• Rewritten experience bullets using XYZ formula
• Rewritten skills section with JD-matched terminology
• List of keywords injected and where
• Warnings about estimated numbers needing verification

Step 4: Diff Generator Agent

Read agents/diff-generator.md and run it with:

• Original resume text
• Tailored resume text (constructed from bullet rewriter output)
• Gap analysis output
• Bullet rewriter output

This agent produces:

• Section-by-section diff of all changes
• Change type classification (rewrite, reorder, add, remove, terminology)
• Reason for each change tied to JD requirements
• Confidence ratings and verification flags
• Summary statistics (X bullets rewritten, Y keywords added, etc.)

Step 5: ATS Scorer Agent

Read agents/ats-scorer.md and run it with:

• Original resume text
• Tailored resume text
• Full JD text

This agent produces:

• Before/after ATS scores with dimension breakdown
• Complete keyword audit table
• Formatting issues
• File format recommendations
• Specific improvement recommendations

Step 6: Synthesize & Output

After all agents complete, merge their outputs into the final deliverable following the Output Format below. Apply these validation rules during synthesis:

Validation Rules:

1. No fabrication — Cross-check every rewritten bullet against the original. If a rewrite adds experience, skills, or achievements not present or inferable from the original, reject it and use a conservative rewrite.
2. Conflict resolution — If the gap analyzer says something is a hard gap but the bullet rewriter added it, flag the conflict. Hard gaps cannot be resolved by rewriting — they must be addressed in the cover letter or action items.
3. Confidence annotations — Any change with low confidence or [verify] tags must appear in the verification checklist.
4. Score consistency — The ATS score improvement should be consistent with the number and type of changes made. A 5-point improvement with 15 keyword additions doesn't add up — investigate.

Gap Only Flow (/resume gap)

1. Structured analysis (Step 1)
2. Gap Analyzer agent (Step 2)
3. Output: Gap report section only (from Output Format)

ATS Only Flow (/resume ats)

1. Structured analysis (Step 1)
2. ATS Scorer agent (Step 5, using original resume only — no "tailored" score)
3. Output: ATS dashboard section only (from Output Format)

Cover Letter Flow (/resume cover)

1. If no prior /resume tailor run in this session, run Steps 1-2 first (need gap analysis)
2. Read agents/cover-writer.md and run it with:
   • Tailored resume (or original if no tailor run)
   • Full JD text
   • Gap analysis output
   • Company name and role title
3. Output: All three cover letter variants (standard, short, talking points)

Interview Prep Flow (/resume interview)

1. If no prior /resume tailor run in this session, run Steps 1-2 first (need gap analysis)
2. Read agents/interview-prep.md and run it with:
   • Gap analysis output
   • Full JD text
   • Tailored resume (or original)
   • Role title and company name
3. Output: Technical questions, gap defense scripts, behavioral questions, questions to ask, red flag preparation

Multi-JD Comparison Flow (/resume compare)

1. Collect 2-5 job descriptions
2. Structured analysis (Step 1) for each JD
3. Read agents/jd-comparator.md and run it with:
   • Full resume text
   • All JD texts with role titles and company names
4. Output: Fit matrix, rankings, shared requirements, skill investment recommendations, application strategy

Output Format

The full /resume tailor output must include ALL of the following sections in this order:

1. Executive Summary

One paragraph containing:

• Overall fit percentage
• Top strength (the single best match between resume and JD)
• Critical gap (the most important thing missing, or "none" if strong match)
• Verdict: one of STRONG_MATCH, STRONG_WITH_GAPS, MODERATE_MATCH, WEAK_MATCH, POOR_FIT
• One-sentence recommendation (apply as-is, tailor and apply, apply with strong cover letter, consider upskilling first)

2. ATS Score Dashboard

ATS COMPATIBILITY
=================
                    Original    Tailored    Change
Keyword Match       ██████░░░░  ████████░░  52% → 85%  (+33)
Formatting          ████████░░  ████████░░  80% → 80%  (+0)
Completeness        ███████░░░  ████████░░  70% → 80%  (+10)
Title Alignment     ████░░░░░░  █████████░  40% → 90%  (+50)
─────────────────────────────────────────────────────────
TOTAL               ██████░░░░  ████████░░  58% → 82%  (+24)

Include the dimension breakdown explanation from the ATS scorer.

3. Gap Report

Ordered by severity:

DEAL-BREAKER gaps (if any):

These gaps may prevent your application from advancing. Address in cover letter and consider upskilling.

SIGNIFICANT gaps:

Important but addressable. Covered by resume tailoring and cover letter.

MINOR gaps:

Low risk. Nice-to-haves or items where your adjacent experience is sufficient.

For each gap: requirement, severity, current status, mitigation strategy.

4. Tailored Resume

The complete tailored resume text, ready to copy-paste. Clean formatting, no JSON, no metadata — just the resume as it should be submitted.

5. Change Log

A table showing every modification:

| # | Section | Change Type | Original | New | Reason | Confidence |
|---|---------|-------------|----------|-----|--------|------------|
| 1 | Summary | add | (none) | "Senior Backend..." | Match JD title | high |
| 2 | Exp - Fooddarzee #1 | rewrite | "Built REST APIs" | "Designed RESTful APIs..." | Add specificity + keywords | medium |

Flag items needing verification with a marker.

6. Verification Checklist

Numbered list of every item the candidate must verify before submitting:

• Estimated numbers that need confirmation
• Reframes that might overstate experience
• Terminology changes that need validation

7. Action Items

Numbered checklist of next steps:

[ ] Review and verify all [verify] tagged items in the tailored resume
[ ] Update resume file with tailored version
[ ] Submit via [recommended format] to [ATS/email]
[ ] Prepare cover letter addressing [critical gap]
[ ] Consider: [skill investment recommendation]
[ ] Run /resume interview for interview preparation

Principles

• Never fabricate experience or skills — reframe what exists, flag what doesn't
• Preserve the candidate's voice — don't over-sanitize into generic corporate speak
• Flag hard gaps honestly — if a requirement genuinely can't be addressed, say so
• Prioritize recency — most relevant/recent experience gets the most attention
• Every change needs a reason — no modifications without a clear JD-driven rationale
• Candidate has final say — the verification checklist exists because the candidate must review and approve every change