Launching today
TraceAgently
See every thought and tool call your AI agent makes. Live.
0 followers
See every thought and tool call your AI agent makes. Live.
0 followers
Your agent breaks. You open the logs. Nothing useful. That's what TraceAgently fixes. A few lines of code gets you a real-time dashboard showing every thought, tool call, tool result, and error in order. Works with OpenAI, Claude, Gemini, LangChain, CrewAI, or any custom loop. What you get: Live trace viewer, Cost per trace calculator, Error pattern detection, Magic Fix (Pro) — AI diagnoses your trace and gives you a fix. Start free with 5,000 events/month. 50% off paid plans for first 3 months.
Hey everyone, builder here.
The thing that pushed me to build this: I had an agent looping 40 times on a single task and had no idea until I got my OpenAI bill.
Standard logging tools just dump text. They don't understand the structure of a reasoning loop. TraceAgently was built specifically for agents, not for web servers.
And if you're building with LangChain or CrewAI, drop your repo in the comments and I'll show you exactly how to plug it in.