One we found this week in a user's repo: a provider answered HTTP 200 with the error buried in the body once you paged too far. Nothing failed, the app just had fewer results than it thought. Curious what yours was and how long it took to find.

Hey PH,

We ve been working with OpenClaw agents and noticed how hard it is to understand what each agent is actually costing in real time.

It s unclear which agent, message, or model is driving spend and costs can quietly spiral.

We re building a simple observability layer to track token usage and costs per agent, with real-time visibility and budget alerts.