Track and analyze AI coding tool usage across your team. Supports multiple providers with a modular architecture.

• Dashboard: Token consumption, costs, model breakdown, top users
• User Analytics: Per-user usage history, model preferences, trends
• Pivot Table: Sortable/filterable view of all users with detailed metrics
• Multi-Provider: Mix and match Claude Code, Cursor, or add your own
• Automated Sync: Cron jobs for continuous data fetching
• CSV Import: Manual import when APIs are unavailable or for backfills

Each provider is optional—configure only the ones you use.

vercel storage create postgres

This automatically sets POSTGRES_URL in your environment.

1. Go to Google Cloud Console → APIs & Services → Credentials
2. Create an OAuth 2.0 Client ID (Web application)
3. Add redirect URI: https://your-app.vercel.app/api/auth/callback/google
4. Copy the Client ID and Client Secret

Set these in Vercel project settings (Settings → Environment Variables):

Add credentials for the providers you want to use:

Redeploy to apply environment variables. Migrations run automatically on build.

npm install # Create .env.local with your credentials cat > .env.local << 'EOF' POSTGRES_URL=postgres://... BETTER_AUTH_SECRET=your-secret-here GOOGLE_CLIENT_ID=your-client-id GOOGLE_CLIENT_SECRET=your-client-secret NEXT_PUBLIC_DOMAIN=yourcompany.com CRON_SECRET=your-cron-secret  # Providers (optional) ANTHROPIC_ADMIN_KEY=sk-admin-... CURSOR_ADMIN_KEY=... EOF npm run cli db:migrate npm run dev

Open http://localhost:3000.

Note: Add http://localhost:3000/api/auth/callback/google to your Google OAuth redirect URIs for local development.

Claude Code usage is tracked via the Anthropic Admin API.

1. Go to Anthropic Console → Settings → Admin API keys
2. Click Create Key (requires org admin access)
3. Copy the key (starts with sk-admin-)

• Schedule: Daily at 6 AM UTC
• Data: Token counts, model, cost, API key ID
• User Mapping: API keys are mapped to emails via the Anthropic API. Unmapped keys appear in the UI for manual assignment.

npm run cli sync anthropic --days 7 npm run cli backfill anthropic --from 2025-01-01 --to 2025-06-01

Cursor usage can be imported via API or CSV.

1. Go to Cursor Team Settings
2. Request admin API access from Cursor support
3. Set CURSOR_ADMIN_KEY in your environment

Sync Behavior:

• Schedule: Hourly
• Data: Token counts, model, user email, cost

npm run cli sync cursor --days 7

Faster for large historical imports.

1. Export from Cursor team dashboard → Usage/Analytics → Export CSV
2. Import:

npm run cli import:cursor-csv /path/to/export.csv

Track AI-attributed commits across your organization by detecting Co-Authored-By headers from Claude, Cursor, Copilot, and other AI tools.

1. Go to GitHub → Your Organization → Settings → Developer settings → GitHub Apps → New GitHub App

2. Configure basic info:

   • GitHub App name: Abacus Commit Tracker (must be unique across GitHub)
   • Homepage URL: Your Abacus deployment URL (e.g., https://abacus.yourcompany.com)

3. Configure webhook:

   • Webhook URL: https://your-app.vercel.app/api/webhooks/github (replace with your actual domain)
   • Webhook secret: Generate with openssl rand -hex 32 and save it for later

4. Set permissions (under "Permissions & events"):

   • Repository permissions:
     • Contents: Read-only (required to fetch commit history)
     • Metadata: Read-only (automatically granted)
   • Subscribe to events:
     • Check Push (this is what triggers the webhook)

5. Configure installation access:

   • Where can this GitHub App be installed?: Select "Only on this account"
   • This restricts the app to your organization only (recommended for internal tools)

6. Click Create GitHub App

7. After creation, on the app settings page:

   • Note the App ID (shown near the top)
   • Scroll down and click Generate a private key (downloads a .pem file)

8. Install the app on your organization:

   • Go to Install App in the left sidebar
   • Click Install next to your organization
   • Choose All repositories or select specific repos
   • After installing, note the Installation ID from the URL: github.com/organizations/ORG/settings/installations/INSTALLATION_ID

# GitHub App credentials (required for production) GITHUB_APP_ID=123456 GITHUB_APP_PRIVATE_KEY="-----BEGIN RSA PRIVATE KEY-----\n...\n-----END RSA PRIVATE KEY-----" GITHUB_APP_INSTALLATION_ID=12345678 GITHUB_WEBHOOK_SECRET=your-webhook-secret # Or for local development/testing (fine-grained token) GITHUB_TOKEN=github_pat_...

Note: The private key must be the full PEM content. In Vercel, you can paste the entire key including newlines, or escape newlines as \n.

For local development, use a fine-grained personal access token (not classic tokens):

1. Go to GitHub → Settings → Developer settings → Fine-grained tokens
2. Generate new token with:
   • Repository access: Select specific repos or "All repositories"
   • Permissions: Contents (read-only)
3. Set GITHUB_TOKEN in .env.local

Fine-grained tokens are scoped to exactly what's needed and don't require org-level access.

• Real-time: Webhook receives push events instantly
• Backfill: Cron job runs every 6 hours to fill gaps (90-day target)
• Detection: Identifies commits with AI attribution patterns

Detected patterns:

npm run cli github:sync getsentry/sentry --dry-run # Test detection (no database needed) npm run cli github:status # Check sync state npm run cli github:sync getsentry/sentry --days 30 # Sync specific repo npm run cli backfill github --from 2024-10-01 # Backfill all repos

# Database npm run cli db:migrate # Run pending migrations # Sync (API-based) npm run cli sync # Sync all providers (last 7 days) npm run cli sync anthropic --days 30 npm run cli sync cursor --days 7 # Backfill (historical data) npm run cli backfill anthropic --from 2025-01-01 --to 2025-06-01 npm run cli backfill cursor --from 2025-01-01 --to 2025-06-01 npm run cli backfill github --from 2024-10-01 # GitHub only needs --from npm run cli backfill:complete anthropic # Mark backfill as complete npm run cli backfill:reset cursor # Reset backfill status # CSV Import npm run cli import:cursor-csv <file> # API Key Mappings (Claude Code) npm run cli mappings # List current mappings npm run cli mappings:sync # Sync from Anthropic API npm run cli mappings:fix # Interactive unmapped key assignment # Data Analysis npm run cli gaps # Check for gaps in usage data npm run cli gaps anthropic npm run cli gaps cursor # Status npm run cli anthropic:status # Show Claude Code sync state npm run cli cursor:status # Show Cursor sync state npm run cli github:status # Show GitHub commits sync state npm run cli stats # Database statistics # GitHub-specific npm run cli github:sync <repo> [--days N] # Sync a specific repo

Vercel cron jobs keep data current:

Note: GitHub commits are primarily synced via webhook (real-time). The backfill cron catches any missed events and fills historical data.

Requires CRON_SECRET to be set.

src/ ├── app/ │ ├── page.tsx # Main dashboard │ ├── users/ # Users list and profiles │ ├── status/ # Sync status page │ └── api/ │ ├── auth/ # Authentication (better-auth) │ └── cron/ # Cron job endpoints ├── lib/ │ ├── queries.ts # Database queries │ ├── sync/ # Provider sync modules │ │ ├── anthropic.ts │ │ ├── cursor.ts │ │ ├── github.ts # GitHub commits tracking │ │ └── index.ts │ └── utils.ts └── scripts/ └── cli.ts # CLI tool 

1. Create src/lib/sync/your-provider.ts implementing the sync interface
2. Add cron routes in src/app/api/cron/
3. Add CLI commands in scripts/cli.ts
4. Update the status page

Each provider should:

• Store data in usage_records with a unique tool identifier
• Aggregate by date/email/model before inserting
• Handle deduplication via the existing upsert logic

Apache 2.0 - see LICENSE