langwatch
Open-Source Agent Observability & Evals for AI Platform teams. Trace, test, route and govern every agent & coding assistant in the company.
What is langwatch?
Open-Source Agent Observability & Evals for AI Platform teams. Trace, test, route and govern every agent & coding assistant in the company.
Overview of langwatch
What it is
LangWatch is an open-source platform for AI in production. It is designed for AI platform teams to trace, test, route and govern every LLM call in a company, including agents built internally and coding assistants used by engineers. The project is licensed under Apache 2.0.
Key features
- LLM Ops: Observability, agent testing, evaluations, prompt management
- Coding Agents: sessions, cost per pull request and per team, privacy controls
- AI Gateway: one OpenAI and Anthropic compatible endpoint, virtual keys, budgets, routing
- AI Governance: every AI tool in the company, who uses it, anomaly rules
Use cases
- Create simulation tests for evaluating or benchmarking agents and voice agents
- Measure the cost per PR and optimize Claude Code usage to make weekly limits last twice as long
- Create virtual keys with budgets to hand out to customers or employees
- Find that 40% of Copilot Studio subscriptions are not actually being used and cancel them
Getting started
To track coding agents:npx langwatch claude # or codex, copilot, opencode, ...To self-host:npx @langwatch/serverOnly Node.js 24 or newer required (macOS or glibc Linux). For production, see Self-hosting LangWatch.
License and pricing
LangWatch is Apache 2.0. The enterprise modules under enterprise/ need a commercial license in production, and the SDKs are MIT. Pricing type is open source.
Based on the project's public README and repository metadata.