GraphOS – Visual runtime and debugger for AI agents (with local-first execution)
GraphOS is an open-source governance and observability tool designed for LangGraph.js to help manage AI agents in production environments. It provides policy enforcement, cost monitoring, and a local-first dashboard with time-travel debugging capabilities. The system runs entirely locally, ensuring no telemetry leaves the user's machine.
- ▪GraphOS prevents infinite loops and runaway costs in AI agent workflows using policy guards like LoopGuard and BudgetGuard.
- ▪It offers a local dashboard with live visualization, session replay, and step-by-step debugging powered by SQLite storage.
- ▪The tool integrates with both TypeScript and Python agents, streaming execution data via WebSocket to a unified interface.
- ▪GraphOS operates without external SaaS, signup, or data telemetry, keeping execution and traces on the user's machine.
- ▪Users can install GraphOS via npm or pip and connect it to their LangGraph applications with minimal code changes.
Opening excerpt (first ~120 words) tap to expand
The Service Mesh for AI Agents. GraphOS is an open-source governance and observability layer for LangGraph.js. Wrap your compiled graph in one line, get policy enforcement (loops, budgets) and a local-first live dashboard with time-travel replay. No SaaS, no signup, no telemetry leaving your machine. 🧐 Why GraphOS? As agents move from demos to production, three things bite: Infinite loops — the agent ping-pongs between nodes, burning tokens silently. Runaway cost — one bad prompt eats your monthly OpenAI budget before you notice. The black-box problem — no way to see what happened inside a 20-step run until it's finished. GraphOS fixes this by wrapping your CompiledGraph with a policy-driven interceptor and streaming every step to a local dashboard.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at GitHub.