What it does
Provides per-agent-pair identity and access management for A2A communication. Policies are written as conditions on a pair of agents, for example allowing an orchestrator agent to call a billing agent only when the amount is below 100. The server exposes tools to define a policy, evaluate a call against policies, and list existing policies.
It suits teams running several agents that call one another and need to restrict which agent can invoke which, and under what conditions. It is one of five MCP servers in the MEOK A2A pack (with prompt injection firewall, data residency, certified handoff, audit logger and rate limiter), and it can also be installed on its own.
Running it means adding the server to an MCP client configuration, such as claude_desktop_config.json for Claude Desktop, or installing the package with pip and launching the agent-policy-enforcement-mcp command over stdio transport. Self-hosting is listed under the MIT licence. Paid tiers are offered by the publisher and add HMAC-signed attestations.
Limitations and requirements
Requires an MCP client configured to launch the server over stdio, or a pip installation of the package. Hosted tiers beyond self-hosting are paid and add attestations verifiable through the publisher's service.
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-policy-enforcement-mcp
Per-agent-pair IAM for A2A. Define policies ('orchestrator may call billing only when amount<100...
v1.0.11As shown on registry.modelcontextprotocol.io when we last read it