Agent MCP Server

MCP server that lets Claude Code agents delegate tasks to agents in other project directories

Local serverstdioPython

What is the Agent MCP server?

Connect Agent to Claude, Cursor or any other MCP client and it stops being a tab you switch to. MCP server that lets Claude Code agents delegate tasks to agents in other project directories. The agent mcp server is what makes that connection.

What the server does

Each agent runs as a separate claude -p session in its own project directory — inheriting that project's MCP servers, CLAUDE.md, and tools. The calling agent just gets the result back.

Available tools

The toolset is worth reading before you wire it up, because it tells you what the integration is really for:

  • list_agents — Lists all configured agents. Call this first to see what's available
  • inspect_agent — Cheap detailed lookup — reads the agent's files without spawning a claude session. Returns the full config (timeout, model, budget, permission mode
  • Groups — A group bundles related agents into a cross-project working set — typically a few code repos plus capability gateways (an infra agent with a
  • dispatch — One-shot task delegation. Results are cached — identical requests within TTL return instantly
  • dispatch_session — Multi-turn: continue a conversation with an agent. First call starts a session, pass session_id back to continue. Never cached
  • dispatch_parallel — Run multiple tasks concurrently. Much faster than sequential dispatch calls
  • dispatch_stream — Same as dispatch but shows live progress while the agent works. Use for long-running tasks. Not cached
  • dispatch_dialogue — Two agents collaborate through multi-turn conversation. Never cached
  • add_agent — Register a new project directory as an agent. Description is auto-generated from project files if omitted
  • update_agent — Update an existing agent's configuration. Only non-empty fields are changed. Pass "none" to clear a field
  • remove_agent — The remove_agent tool exposed by this server

Installation

agent-dispatch on PyPI is all you need. Most clients run it directly, so configuration is a few lines and a restart.

Where it fits

Among the developer tooling options, the useful question is rarely "what can it do" but "what does it cost you to run" — permissions, credentials, and how much of your context its toolset consumes. Agent's toolset — list_agents, inspect_agent, Groups and 8 more — is a fair guide to whether it matches your workflow. It is maintained by ginkida; worth a glance at recent repository activity before you build anything load-bearing on it.

SyncDev reviews every entry in this directory against the project's own documentation before publishing, and revisits them as servers change.

Worth knowing first

  • It runs with your machine's permissions. That is convenient and also the reason to think about what you point it at before you approve a tool call.
  • With 11 tools registered it takes up a noticeable share of the context window; turn it off in projects that never touch Agent.
  • Keep per-call confirmation enabled while you learn its behaviour; it is the cheapest safeguard you have.

Available tools

ToolWhat it does
list_agentsLists all configured agents. **Call this first** to see what's available.
inspect_agentCheap detailed lookup — reads the agent's files without spawning a claude session. Returns the full config (timeout, model, budget, permission mode, allowed/disallowed tools), detected MCP/stacks/DBs, plus short previews
GroupsA **group** bundles related agents into a cross-project working set — typically a few code repos plus capability gateways (an infra agent with a Portainer MCP, an analytics agent with a browser + Yandex Metrica). It lets
dispatchOne-shot task delegation. Results are cached — identical requests within TTL return instantly.
dispatch_sessionMulti-turn: continue a conversation with an agent. First call starts a session, pass session_id back to continue. Never cached.
dispatch_parallelRun multiple tasks concurrently. Much faster than sequential dispatch calls.
dispatch_streamSame as dispatch but shows live progress while the agent works. Use for long-running tasks. Not cached.
dispatch_dialogueTwo agents collaborate through multi-turn conversation. Never cached.
add_agentRegister a new project directory as an agent. Description is auto-generated from project files if omitted.
update_agentUpdate an existing agent's configuration. Only non-empty fields are changed. Pass "none" to clear a field.
remove_agentThe remove_agent tool exposed by this server.

How to install the Agent MCP server

{
  "mcpServers": {
    "agent-dispatch": {
      "command": "uvx",
      "args": ["agent-dispatch"]
    }
  }
}

Add to claude_desktop_config.json, then restart Claude Desktop.

Example prompts to try

  • Use Agent to list agents.
  • Use Agent to inspect agent.
  • Use Agent to Groups.

Frequently asked questions

It connects Agent to MCP-compatible AI assistants such as Claude and Cursor, exposing 11 tools (list_agents, inspect_agent, Groups, and more) that the assistant can call on your behalf. Instead of copying data back and forth by hand, the assistant works with Agent directly.