Skip to content

Getting Started

Installation

go install github.com/xrsl/cvx@latest

Prerequisites

Setup

Run the initialization wizard in your repository:

cvx init

The wizard will:

  1. Link your GitHub repo - auto-detected from git remote
  2. Select CLI agent - Claude or Gemini
  3. Set CV source path - your CV data file
  4. Set letter source path - your letter data file
  5. Set reference directory - additional context for AI
  6. Create/link GitHub Project - with job-tracking statuses

Your First Application

1. Add a Job Posting

cvx add https://company.com/careers/software-engineer

This will:

  1. Fetch and parse the job posting
  2. Extract structured details using AI
  3. Create a GitHub issue
  4. Add to your project board

2. Analyze Match Quality

cvx advise 42

Get AI-powered analysis of how well your CV matches the position.

3. Build Tailored Documents

Interactive Mode (default):

cvx build 42

Uses Claude or Gemini CLI for real-time editing with tool use.

Agent Mode:

cvx build -m sonnet-4

Structured TOML output with Pydantic validation.

4. Approve and Submit

cvx approve 42

Commits, tags, pushes, and updates project status to "Applied".

5. View Later

cvx view 42

Opens the PDF from the git tag.

Workflow Summary

cvx init                    # Initialize project
cvx add <url>               # Add job posting
cvx advise <issue>          # Analyze job-CV match
cvx build [issue]           # Build tailored CV/cover letter
cvx approve [issue]         # Commit, tag, push
cvx view <issue>            # View submitted documents

Build Modes

Interactive CLI Mode (Default)

Best for iterative refinement:

cvx build 42                # Start/resume session
cvx build -c "more ML focus" # Add feedback

Benefits:

  • Direct file editing by AI
  • Session persistence
  • Interactive conversation
  • Auto-detects claude or gemini CLI

Agent Mode

Best for structured, repeatable output:

cvx build -m sonnet-4       # Claude
cvx build -m flash-2-5      # Gemini
cvx build -m qwen3-32b      # Groq

Benefits:

  • Schema-validated output
  • Multi-provider support
  • Automatic fallback on failure
  • TOML output

Requirements:

  • uv installed
  • Data files: src/cv.toml, src/letter.toml
  • Schema: schema/schema.json

Other Commands

cvx list                    # View all applications
cvx list --company google   # Filter by company
cvx rm <issue>              # Remove application

Next Steps