Pulse — Session Telemetry for Claude Code

You deployed Claude Code to your team. Now answer:

Three questions every engineering leader asks after rolling out AI-assisted development.

How much are we spending?

Track cost per session, per developer, per project. See daily trends and budget impact at a glance.

Is it actually producing code?

Correlate AI session time with git output: commits, files changed, lines added and removed.

Who is using it?

Developer activity breakdown with session counts, hours, and machine-level detail across your org.

What Pulse tracks

Cost per session

USD cost tracked with precision(10,4) on every session. Aggregated per developer, project, and org with trend indicators.

Git output correlation

Commits, files changed, lines added/removed captured at session end. See which sessions produce real code output.

Dashboard with charts

KPI cards with trend arrows, daily activity charts, top developer rankings. Auto-refresh for live monitoring.

Pilot verification

Built-in pilot form for beta testers to verify their setup. Admin review with approve/flag workflow.

Quick start

      # Clone and configure

      git clone https://github.com/vamadeus-dev/pulse.git

      cd pulse

      cp .env.example .env

      # Edit .env with your API_KEY and PUBLIC_URL

      # Start with Docker

      docker compose up -d

      # Open the dashboard

      open http://localhost:3100/admin

🔒 Self-hosted — your data stays on your infrastructure

📦 Docker Compose — one command to deploy

🛠 MIT License — use it however you want