CVFormatterAgent — AI skill for Claude Code

CV Formatter Agent

An AI-powered agent that reformats your CV to maximise compatibility with Applicant Tracking Systems (ATS). Built on the [Strands Agents SDK](https://strandsagents.com/) with Amazon Bedrock (Claude Sonnet 4) as the default LLM.

What it does

1. Parses your CV from PDF, DOCX, or TXT format.
2. Extracts required and preferred keywords from a job specification.
3. Analyses keyword gaps between your CV and the job spec, computing an ATS score.
4. Reformats your CV to ATS standards: standard section headers, single-column plain text, MM/YYYY dates, Skills immediately after Summary.
5. Generates a .txt and a .docx output file, each with a keyword match report appended.

Requirements

• Python 3.11+
• AWS credentials configured (for Amazon Bedrock), or set MODEL_PROVIDER / MODEL_ID env vars to use a different provider.

Installation

pip install -e ".[dev]"

Or install runtime dependencies only:

pip install strands-agents>=1.0.0 pdfplumber>=0.11.0 python-docx>=1.1.0

Usage

python main.py path/to/your_cv.pdf path/to/job_spec.txt

Or pass the job spec as an inline string:

python main.py your_cv.docx "We are looking for a Python developer with 5+ years experience..."

Output files are written to the `output/` directory by default:

output/
  your_cv.txt     # ATS-optimised plain text CV + keyword report
  your_cv.docx    # ATS-optimised DOCX CV + keyword report

Environment variables

Variable	Default	Description
MODEL_PROVIDER	bedrock	LLM provider (bedrock, anthropic, etc.)
MODEL_ID	us.anthropic.claude-sonnet-4-20250514-v1:0	Model identifier

Project structure

cv-formatter-agent/
  main.py                         # CLI entry point
  agent.py                        # Strands Agent configuration
  models.py                       # Shared dataclass models
  exceptions.py                   #