Show HN: AgentTrafficCop – Stop AI agents from looping and burning money

AgentTrafficCop launched as an open-source, self-hostable runtime circuit breaker for multi-agent systems that validates every agent handoff against a strict Pydantic v2 contract and uses SHA-256 payload fingerprinting plus step counters to trip before a loop exhausts an API budget. The tool runs as a standalone FastAPI service, works with LangGraph, CrewAI, AutoGen or raw asyncio, and ships a LangGraph adapter that wraps any node in one line; when thresholds breach, it returns structured diagnostics so the orchestrator can halt, alert or escalate. The middleware and SDK are open source today, with managed tiers announced as coming soon.

AgentTrafficCop is a runtime circuit breaker for multi-agent systems. It validates every handoff, fingerprints every payload, and trips before a loop empties your API budget. Every handoff validated against a strict Pydantic v2 contract before it reaches the next agent. Malformed payloads are rejected at the door. SHA-256 fingerprinting on every payload. Identical repeats and A↔B ping-pongs trip the breaker. Step counters catch the rest. When thresholds breach, the response carries structured diagnostics. Your orchestrator decides: halt, alert, or escalate. python your orchestrator.py from agenttrafficcop import Guard, LoopDetected guard = Guard url="https://api.agenttrafficcop.com", api key="your-key" try: result = await guard.check session id="run-42", sender="coder", recipient="reviewer", payload={"diff": "fix typo"}, print "approved at step", result.step except LoopDetected as e: print "loop caught:", e.error code Runs as a standalone FastAPI service. Works with LangGraph, CrewAI, AutoGen, or raw asyncio. The SDK ships a LangGraph adapter — wrap any node in one line. The middleware and SDK are open source and self-hostable today. Managed tiers are coming soon.