Skip to content

MCP Tool Access Control is a Content Security: Content Security Providers govern the MCP tool traffic flowing through Aembit's MCP Identity Gateway. You add a provider to the Content Security component of an Access Policy.Learn more that governs which Model Context Protocol: A standard protocol for AI agent and server interactions that defines how AI assistants communicate with external tools and data sources.Learn more(opens in new tab) tools an AI agent can discover and which it can call.

Authorizing an MCP Server: A server that implements the Model Context Protocol to provide tools, resources, or data to AI agents and MCP clients.Learn more(opens in new tab) grants an AI agent access to every MCP tool that the MCP server exposes. MCP Tool Access Control lets you control that access. You choose which MCP tools to allow and which to block. The MCP Identity Gateway: A component that brokers MCP traffic between MCP clients and target MCP servers, validating authorization and presenting Aembit-managed credentials on each request.Learn more enforces your choices on every request your Access Policy: Access Policies define, enforce, and audit access between Client and Server Workloads by cryptographically verifying workload identity and contextual factors rather than relying on static secrets.Learn more governs.

An Access Policy can carry both MCP Tool Access Control and CrowdStrike AIDR: CrowdStrike AI Detection and Response (AIDR) is a content inspection service that scans MCP tool listings, tool call inputs, and tool call outputs in real time and returns an allow, block, or transform verdict to Aembit.Learn more. MCP Tool Access Control matches tool names first, so AIDR inspects only the MCP tools and calls that MCP Tool Access Control allowed.

To add MCP Tool Access Control to an Access Policy, see Add MCP Tool Access Control to an Access Policy.

MCP Tool Access Control governs how AI agents discover MCP tools and how they call them. You add the MCP tools you want to govern, then allow or block them.

  • The Tool Visibility: A control that governs which MCP tools an agent can discover when it requests a tool listing.Learn more control decides which MCP tools an AI agent discovers when it asks an MCP server what MCP tools exist. A blocked MCP tool never appears in that listing.
  • The Tool Invocation: A control that governs which MCP tools an agent can call, independently of which tools it can discover.Learn more control decides which MCP tools an AI agent can call. The MCP Identity Gateway returns an error for a call to a blocked tool and never forwards it to the MCP server.

Both controls act on standard MCP methods, tools/list and tools/call. See Proxied MCP methods.

Aembit records what each control decided in the Workload Event: A structured audit record that Aembit generates when application-level traffic flows through Agent Proxy or MCP Identity Gateway.Learn more, so an event shows why an AI agent didn’t discover an MCP tool or why a call failed.

Aembit doesn’t collect or store the MCP tool list from an MCP server, so you need to know the MCP tool names you want to govern. Take them from the MCP server’s documentation, which lists the names it publishes, such as get_issue.

You can add an exact name or a glob pattern, and matching is case-sensitive. For the full syntax and the field limits, see MCP Tool Access Control tool name reference.