What it does
Most MCP servers rate-limit on their own, so an agent calling several of them can exceed the intended quota across the set. This server acts as a common counter: each MCP server checks it before allowing a call, in A2A and multi-MCP deployments. Enforcement uses a sliding window and concurrency grants, and the publisher describes signed enforcement attestations.
It suits teams running several MCP servers or agent-to-agent setups that need one quota shared across them. It exposes three tools, acquire, release and status, which an assistant can call directly. The publisher describes it as part of a fleet of MCP servers from MEOK AI Labs and lists its license as MIT.
To run it, add the server to an MCP client configuration, such as claude_desktop_config.json for Claude Desktop. Alternatively, install the Python package with pip and start the agent-rate-limiter-mcp command, which uses stdio transport.
Limitations and requirements
Signed enforcement attestations are added only in paid tiers above self-hosting, and are verified at verify.meok.ai.
Safety
Our team has not reviewed this resource. Only automated rules checked it; resources with high-risk signals are not published. These are signals, not a guarantee that installing it is safe. Try it in a sandbox project first.
- Runs code or changes an assistant's behaviour
Sources
- Official MCP Registry first seen 13 Nov 2026 · last seen 13 Nov 2026 · its page on the source
io.github.CSOAI-ORG/agent-rate-limiter-mcp
Fleet-wide shared rate limiter for A2A + multi-MCP deployments. Most MCP servers rate-limit inde...
v1.0.10As shown on registry.modelcontextprotocol.io when we last read it