io.github.CSOAI-ORG/agent-rate-limiter-mcp
Fleet-wide shared rate limiter for A2A + multi-MCP deployments. Most MCP servers rate-limit inde...
Versions
1.0.10latest1.0.2Tools 6
acquire Request a rate-limit slot for an agent action. Returns {allowed, current, limit, retry_after_sec}. - tenant_id: the customer / organisation identifier - agent_id: which agent is calling (may be a chain: "orchestrator:supervisor:worker") - operation: a key to scope counters (e.g. "llm_call", "db_write", "external_api") - window_sec: sliding window in seconds (default 60) - weight: how many tokens this call consumes (default 1 — expensive calls can weight >1)
release Release a rate-limit slot. Doesn't refund tokens in the sliding window (those naturally expire), but removes the grant from the active-calls set — useful for concurrency-limit tools that track in-flight operations.
status Inspect usage. If agent_id is empty, returns tenant-wide summary.
set_quota Override the default quota for a (tenant, agent, operation) tuple. Pro+ only.
reset_counters Clear counters for a tenant (or a specific agent within a tenant). Pro+ only. Useful for daily resets, test harnesses, emergency overrides.
sign_rate_limit_attestation Emit a cryptographically signed attestation of rate-limit enforcement over a window. Enterprise audit evidence — show regulators you were enforcing quotas during the incident window.
Permissions 3
env_vars low network medium filesystem low