Your AI agent needs a brake pedal: stopping loops and runaway cost at runtime

Vijay M, a student founder in Bengaluru, has released SteerPlane, an open-source runtime guard for AI agents that enforces cost, step and denied-action limits outside the model. The library wraps agent functions with a guard decorator (e.g. max_cost_usd=5, denied_actions=["delete_*"]) and halts a run when it exceeds budget or attempts a blocked action, addressing failures such as a coding agent that edited 48,218 files in 103 seconds and a production database and backups deleted in 9 seconds.

AI agents fail in boring, expensive ways. They repeat the same step all night. They run up an API bill. They delete something nobody told them to touch. The public incident reports from 2026 are already piling up: a coding agent that edited 48,218 files in 103 seconds, and a production database and its backups gone in 9 seconds. A prompt that says "be careful" does not fix this. The rule has to live in code, outside the model, where the agent cannot talk its way past it. SteerPlane runs next to the agent and enforces limits while it runs: python from steerplane import guard @guard max cost usd=5, denied actions= "delete " def run agent task : ... If the agent goes over budget or tries a denied action, the run stops. You can also set a step limit with max steps. It does not filter content or judge whether an answer is true. It limits what an agent can do and spend. pip install steerplane Repo: https://github.com/vijaym2k6/SteerPlane https://github.com/vijaym2k6/SteerPlane Site: https://www.steerplane.com https://www.steerplane.com SteerPlane is open source. Patent Published. I'm Vijay M, a student founder in Bengaluru. I'd like to hear which failure has hurt you most: runaway cost, loops, or bad actions? The safety layer for the AGI Era.