mcpdirectory.dev
9,129 servers · updated 23 Sept

Acp MCP server

official from Agenticcontrolplane Active last commit 20 d 
{
  "mcpServers": {
    "agenticcontrolplane-acp": {
      "type": "http",
      "url": "https://api.agenticcontrolplane.com/mcp"
    }
  }
}
file .mcp.json (in your project root)version 0.3.0

What it does

This MCP server integrates Claude, ChatGPT, Cursor, and other AI clients with Agentic Control Plane (ACP) to enforce governance policies on tool calls before execution. It routes tool-call requests through ACP's policy engine, which returns allow/deny/ask decisions and logs audit trails tied to the human user identity.

  • Check whether an AI agent should be allowed to execute a sensitive tool call (like database writes or API access) before it runs.
  • Maintain a complete audit log of every tool call made by AI agents across your team, attributable to the human who initiated it.
  • Implement rate limits, scope checks, and delegation policies on tool use without modifying your agent or client code.

Tools this server exposes

ToolDoes
acp_checkAsk ACP whether a tool call is allowed; returns allow, deny, or ask (for confirmation) with a reason.
acp_statusVerify the connection to ACP and confirm your workspace identity.

Before you enable it

An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.