Resume Tailor Skill Documentation

Resume Tailor is an Claude Code skill that turns a job description and a candidate's master profile into ATS-verified, tailored resumes and matching cover letters compiled into publication-quality PDFs.

How to Use It

1. Download the latest release

Download the latest version from the Releases page.

2. Install the skill

Go to your Claude Code skills directory:

cd ~/.claude/skills

Then extract the resume-tailor.skill file into this directory.

3. Add your resume and cover letter

Place your master resume and cover letter in the ~/cv directory:

~/cv/
├── master-profile.md
└── cover-letter.md

You can configure the skill to use a different directory if needed.

4. Tailor your resume

From Claude Code, simply run:

/resume-tailor <link_to_job>

For example:

/resume-tailor https://www.linkedin.com/jobs/view/123456789

The skill will analyze the job description, tailor your resume and cover letter, and generate the final output.

5. Find your tailored resume

The generated files will be organized by role, company, and job ID:

~/cv/<role>/<company_name>/<job_id>/

For example:

~/cv/software-engineer/acme/123456789/

Key Principles & Core Philosophy

Code Decides Quality, LLM Writes Content
The LLM generates resume content, but deterministic code controls layout constraints, character counts, ATS extraction, and PDF visual bounds. Page count and text alignment are verified against the compiled PDF—never estimated.
Strict Schema Guardrails
Layout overflow is prevented at the schema level. Pydantic schema validation enforces bullet point counts (e.g., maximum 4 per role) and length limits (e.g., maximum 120 characters per bullet).
The Honesty Constraint
- Never fabricate skills or metrics: If a skill is missing from the master profile, it is categorized as a genuine gap.
- Translate adjacent experience: Existing experience is reframed using the job description's vocabulary (e.g., reframing a legacy migration project as scalable architecture).
- Traceable numbers: All metrics and statistics must originate directly from the master profile.
Human-in-the-Loop Verification
Proposed reframings and gap fillings must be explicitly reviewed and approved by the user before resume compilation.

Execution Workflow Diagram

flowchart TD
    subgraph Input ["1. Inputs & Preflight"]
        A1[Master Profile\nmaster_profile.md]
        A2[Job Description / URL]
        A3[Preflight Check\npreflight.py]
    end

    subgraph Analysis ["2. JD Parsing & Gap Analysis"]
        A2 --> B1[Extract Technical Requirements\n& Keywords]
        B1 --> B2[Gap Analysis vs Master Profile\n- Direct Hits\n- Bridgeable Experience\n- Genuine Gaps]
        A1 --> B2
    end

    subgraph Approval ["3. Human Verification"]
        B2 --> C1{User Approval\n& Edits}
        C1 -- Rejected / Modifications --> B2
    end

    subgraph SchemaRender ["4. Schema & Compilation Loop"]
        C1 -- Approved --> D1[Structured JSON Payload\nresume_data.json]
        D1 --> D2[Pydantic Schema Guardrails\n& Style Linting schema.py]
        D2 --> D3[LaTeX Sanitization\n& Jinja2 Rendering render.py]
        D3 --> D4[LaTeX Compilation\npdflatex]
        D4 --> D5{Autofit & Page Bounds OK?}
        D5 -- Bounds Exhausted / Overflow --> D1
    end

    subgraph Verification ["5. Verification & Output"]
        D5 -- Pass --> E1[ATS Text Extraction Check\nverify.py]
        E1 --> E2{ATS Rules Pass?}
        E2 -- Fail --> D1
        E2 -- Pass --> E3[Render Page JPEGs\npdftoppm]
        E3 --> E4[Visual Verification & Artifact Packaging\n$CV_HOME/role/company/job-id/]
    end

    subgraph CoverLetter ["6. Parallel Cover Letter Track"]
        C1 -- Approved --> F1[Company Culture & Value Research]
        F1 --> F2[Generate Cover Letter JSON\ncover_data.json]
        F2 --> F3[Sanitize & Compile PDF\ncover_letter.tex.j2]
        F3 --> E4
    end

Prerequisites & Environment Setup

Before executing the skill, system dependencies must be validated via the preflight script:

python3 scripts/preflight.py

System Dependencies

CLI Tools: pdflatex (TeX Live / MacTeX), pdftoppm (Poppler utilities)
Python Libraries: pydantic, jinja2, pymupdf (PyMuPDF / fitz), python-docx (optional for Word export)
User Files:
- master_profile.md: Single source of truth for candidate background (see assets/master_profile.example.md).
- $CV_HOME: Destination environment variable (defaults to ~/cv).

Detailed Step-by-Step Execution

1. Job Description Extraction

Fetches and stores the job posting verbatim as job-description.md inside the application output folder.
Extracts technical keywords, preferred skills, core responsibilities, and seniority markers.

2. Gap Analysis & Technical Alignment

Compares extracted requirements against the master profile.
Produces three distinct categorizations:
- Direct Hits: Exact skill matches from the candidate profile.
- Bridgeable Reframings: Relevant adjacent experience mapped into the target role's terminology.
- Genuine Gaps: Requirements missing from candidate history.

3. Human Verification Step

Pauses execution to present proposed bullet points, vocabulary alignment, and reframings to the user.
Requires explicit user approval before proceeding to artifact generation.

4. Schema Generation & Style Linting

Constructs resume_data.json bound by constraints in scripts/schema.py.
Guardrails: Max 4 bullets per role, max 120 characters per bullet point.
Style Rules (lint_style()): Enforces active verbs, blocks passive wording, em-dashes, or fluff adjectives.

5. LaTeX Sanitization & Rendering

Passes raw payload through LaTeX escaping (% $\rightarrow$ \%, & $\rightarrow$ \&, $ $\rightarrow$ \$, etc.).
Inject values into Jinja2 templates (assets/resume.tex.j2).
Runs Autofit Loop: Compiles with pdflatex iteratively adjusting typography density (font size, margins, spacing) to guarantee exact page length without dropping content.

6. Automated ATS & Layout Verification

Uses scripts/verify.py to inspect the compiled PDF:
- Verifies raw text extraction via PyMuPDF.
- Ensures section headers and chronological flow are preserved.
- Verifies employer and date proximity (dates_stay_with_employer).
Rejects build with non-zero exit code if layout or text extraction criteria fail.

7. Visual Inspection & Artifact Directory

Renders high-resolution JPEG images of compiled PDF pages using pdftoppm.
Stores final outputs under structured path: $CV_HOME/<role>/<company>/<job-id>/
- resume.pdf & resume.tex
- resume_data.json & job-description.md
- page-1.jpg, page-2.jpg
- build-report.json

8. Parallel Cover Letter Track

Conducts targeted research on company culture and engineering values.
Synthesizes technical cover letter framework into cover_data.json.
Compiles companion PDF using assets/cover_letter.tex.j2 through the same sanitization pipeline.

Directory Structure & Script Index

.
├── SKILL.md                          # Main Skill Definition & Instruction Set
├── README.md                         # Architecture & Operational Documentation
├── assets/
│   ├── master_profile.example.md     # Example Master Profile Schema
│   ├── resume.tex.j2                 # Jinja2 LaTeX Template for Resumes
│   └── cover_letter.tex.j2           # Jinja2 LaTeX Template for Cover Letters
├── references/
│   ├── workflow.md                   # In-depth JD Extraction & Alignment Guide
│   └── troubleshooting.md            # Resolution guide for verification errors
└── scripts/
    ├── preflight.py                  # Environment & dependency check
    ├── build.py                      # Main orchestrator entry point
    ├── schema.py                     # Pydantic schemas, escaping & linting
    ├── render.py                     # LaTeX rendering & autofit compilation
    ├── render_docx.py                # Optional DOCX exporter
    └── verify.py                     # Automated ATS & layout verification tool

Command Reference

Preflight Check

python3 scripts/preflight.py

Full Application Build

python3 scripts/build.py resume_data.json \
  --role ml-engineer \
  --company stripe \
  --job-id 4821 \
  --jd job-description.md \
  --cover cover_data.json

Standalone PDF Audit

python3 scripts/verify.py resume.pdf

Design Options: Date Alignment Trade-off

The skill supports two date formatting options in resume_data.json:

"date_style": "right" (Default)
Flush-right dates using \hfill. Visually clean for human readers, though wide whitespace gaps may cause some raw text extractors to separate dates from employers.
"date_style": "inline"
Formats headers as Company | Location | Dates. Guarantees employer and date remain adjacent during ATS text extraction. Recommended for automated enterprise ATS screening.

Claude Resume Generator Skill

At a glance

README