Token Police

Token Police is an SDK (Node, Python) that sits in front of every LLM call your app makes. Before the call goes out it checks your rules — "block this user after $5 a month", "tell me when AI Search passes $100 a week", "send free users to a cheaper model" — and lets it through, reroutes it, or refuses it. It runs in your process, not as a gateway: prompt and completion text never leave your servers. It fails open, so if the service is unreachable your call proceeds untouched. Dry-run is the default: nothing is blocked until you say so, and the first thing you see is a report of what each rule would have blocked, per user, with the dollar amount. Loop detection catches an agent repeating the same call inside one run. Two lines to install. Free while we launch.