About this project
TokenOps is a Python library and control plane that brings token governance to multi-agent AI workflows. It sets a single budget for an entire agent run and enforces it before every model call, stopping the run when the budget is exceeded. It features a shared ledger across processes, ten built-in policies, and a dashboard for cost attribution per agent. It integrates with agents via a simple wrapper around the completion function, supports FastAPI/A2A services, and can be deployed with Docker or make. It is not a gateway or tracing tool but complements them.
Comments
0 people shared their preference · Deer Point appears after 10 participants
Sign in to join the discussion.