Skip to content

MCP governance

An MCP server hands an agent a list of tools and lets it call them. Exposed as it comes, every caller sees every tool, tool names leak whatever the backend happens to call them, and one agent's traffic is indistinguishable from another's. Governing an MCP proxy decides who connects, which tools they see, what those tools are called, and how often they can be invoked.

Where MCP policies attach

These policies attach to an Mcp resource, so they cover the MCP traffic that proxy carries.

Authentication and authorization run before a call reaches the server: the first establishes who is calling, the second checks that this caller may use the specific tool, resource, or prompt named in the request. Access control and rewriting shape the list the caller sees in the first place, so a tool that is filtered out is never offered rather than refused on use. Rate limiting counts calls per tool, resource, prompt, or JSON-RPC method, which is a finer unit than a request count.

MCP policies

These policies are documented in the Policy Hub, the versioned reference for every API Platform policy. For policy categories and how policies chain, see the Policy Hub overview.

Policy What it does
MCP Authentication Secures MCP server traffic per the MCP specification authorization profile
MCP Authorization Validates access to MCP tools, resources, and prompts using JWT claims or OAuth scopes
MCP Access Control Controls which tools, resources, and prompts a caller can reach using allow/deny lists
MCP Rewrite Defines user-facing tool names and maps them to backend capability names
MCP Rate Limit Applies rate limits to MCP traffic per tool, resource, prompt, or JSON-RPC method
Semantic Tool Filtering Filters MCP tools to only those semantically relevant to the user query