HireGraph

Your code is your resume. Everything runs on your machine.

HireGraph is a local-first CLI that turns your code into job applications. It scans your projects, builds a skill graph, matches you against jobs from public ATS APIs, generates tailored resumes, and submits applications — all from your terminal.

Open Source | Local-First | Bring Your Own Key

Install

npm install -g hiregraph

For Claude Code users (recommended)

npm install -g hiregraph
hiregraph install-skill

Then open Claude Code and say: "Set up hiregraph and scan my projects". Claude Code handles everything — no manual commands needed.

Or install the skill directly

npx skills add https://github.com/c0ncepT23/hiregraph --skill hiregraph --yes --global

API Key

HireGraph reads ANTHROPIC_API_KEY from your environment. Set it once:

• Mac/Linux: export ANTHROPIC_API_KEY="sk-ant-..." (add to ~/.bashrc or ~/.zshrc)
• Windows: Add ANTHROPIC_API_KEY to System Environment Variables
• Claude Code: Already set automatically

Quick Start

# 1. Set up your profile
hiregraph init

# 2. Scan your projects
hiregraph scan ~/projects/my-app
hiregraph scan ~/projects/another-app

# 3. Check your skill graph
hiregraph status

# 4. Fetch jobs from Greenhouse, Lever, and Ashby
hiregraph jobs

# 5. Find your best matches
hiregraph matches

# 6. Apply with a tailored resume
hiregraph apply gh_12345 --review

# 7. Track your applications
hiregraph history

How It Works

Scan: 7 Layers of Code Analysis

When you run hiregraph scan, seven layers extract structured data from your codebase:

| Layer | What It Does | LLM? | |-------|-------------|------| | 1. File Discovery | Walk file tree, count LOC, detect languages | No | | 2. Dependencies | Parse package.json, requirements.txt, Cargo.toml, go.mod | No | | 3. AST Analysis | Analyze code structure — functions, classes, components, hooks | No | | 4. Git Forensics | Commits, active days, contributors, velocity | No | | 5. Quality Signals | Test ratio, complexity, type safety, secrets scan | No | | 6. Architecture | Detect patterns — Service Layer, MVC, Repository, Monorepo | No | | 7. Classification | One Haiku call on a structured summary (not your code) | Yes |

Layers 1-6 are completely local. No network calls. No code leaves your machine.

Match: Two-Tier Matching

1. Pre-filter — TF-IDF vector similarity finds the top 50 candidates from thousands of jobs. Runs locally in milliseconds.
2. LLM evaluation — Haiku scores each candidate with reasoning, strengths, and gaps. ~$0.15 for 50 evaluations.

Apply: Tailored Resume + Auto-Submit

• One Haiku call tailors your summary and selects relevant project bullets per job
• pdfkit generates an ATS-compatible PDF locally
• Submits through official ATS APIs (same ones Indeed/LinkedIn use)
• The company sees a normal application — they have no idea HireGraph exists

Commands

| Command | What It Does | |---------|-------------| | hiregraph init | Set up your profile (resume upload + preferences) | | hiregraph scan <path> | Scan a project and update your skill graph | | hiregraph status | Show your skill graph summary | | hiregraph jobs | Fetch jobs from Greenhouse, Lever, and Ashby | | hiregraph matches | Match your skill graph against fetched jobs | | hiregraph apply <job-id> | Generate tailored resume and submit | | hiregraph history | View and manage application history |

Apply Options

hiregraph apply gh_12345              # Apply to a specific job
hiregraph apply gh_12345 --review     # Preview resume before submitting
hiregraph apply gh_12345 --dry-run    # Generate PDF without submitting
hiregraph apply --all-above 8         # Batch apply to all 8+ matches

History Options

hiregraph history                              # List all applications
hiregraph history update app_x9y8 --status interview
hiregraph history update app_x9y8 --status offer --notes "Start date March"

What's On Your Machine

~/.hiregraph/
  identity.json          # Your profile
  skill-graph.json       # Accumulated skill graph from scans
  config.json            # Preferences
  jobs/                  # Cached job listings
  matches/               # Match results with scores
  resumes/               # Generated PDFs
  history.json           # Application tracking

Everything is a file. You can inspect it, back it up, version control it, or delete it.

What It Costs

The only cost is LLM API usage on your existing Anthropic key:

| Action | Cost | |--------|------| | Scan a project | ~$0.003 (1 Haiku call) | | Parse ~5,000 job descriptions (first run) | ~$1.50 (cached after) | | Match (50 evaluations) | ~$0.15 | | Tailor + apply per job | ~$0.003 | | Total first run | ~$1.70 | | Daily refresh + match | ~$0.15 |

Supported ATS

| ATS | Fetch Jobs | Auto-Submit | |-----|-----------|-------------| | Greenhouse | Yes | Yes | | Lever | Yes | Yes | | Ashby | Yes | Yes |

63 companies included in the seed registry. Add your own to ~/.hiregraph/companies.json.

Requirements

• Node.js >= 18
• ANTHROPIC_API_KEY in your environment (auto-set by Claude Code / Cursor)

License

MIT