Algorithms
Detailed implementation of each rate-limiting and enforcement algorithm.
Token Bucket
Continuous-time token bucket rate limiter — algorithm, configuration, and state format.
Cost-Based Budget
Fixed-period quota limiter with staged warn/throttle/reject actions.
LLM Token Limiter
TPM/TPD token limiter with pessimistic reservation, reconciliation, and optional streaming enforcement.
Loop Detector
Sliding-window duplicate request detector for agentic loop protection.
Kill Switch
Descriptor-based emergency blocker evaluated before any rate-limit rule.