MCP can quietly burn through your AI budget.

One thing we’ve been thinking about while building Traccia: an agent can get stuck making repeated tool calls, or keep switching to expensive models, without actually making meaningful progress.

And the scary part is that the run may still technically succeed.

That’s why we think agent policies need to include cost and execution -

runtime policy enforcement:

→ Limit tool calls per run
→ Restrict which models can be called
→ Set execution/time limits
→ Stop the run when it crosses a defined boundary

Thats exactly the role an AI Agent Control Plane plays.

How are you controlling agent costs today?