Agent Intern
Delegate from Claude Code to Gemini, Codex, Copilot, Cursor and opencode CLIs as sub-agents.
At a glance
Not tested yet- GitHub stars
- 34
- Weekly downloads
- 1,432
- Last commit
- 2026-11-18
- Licence
- MIT
- Transport
- stdio
- Auth
- none
uvx agent-intern
What it does
Agent Intern is an MCP server that runs coding CLIs you are already signed in to as headless sub-agents of Claude Code or Antigravity. Claude stays in charge and the intern runs the errand under your own login, so it spends that CLI's plan instead of your Claude quota. It is for people who have Gemini, ChatGPT, Copilot or Cursor plans sitting idle while Claude Code runs out of quota. It also works the other way, adding a Claude Code backend to hosts that are not Claude Code.

Tools it gives your agent
opencode_status: checks the opencode backend and spends no quota.codex_ask: hands work to Codex.claude_status: answers in Antigravity, Copilot, Cursor and opencode.claude_ask: hands work to Claude on your own Claude login.agent_swarm: runs tasks in parallel across workers and returns one answer block.agent_stats: shows calls per backend, error rates, recent swarms and your Antigravity quota.
Quick start
- Install uv, which runs the server.
- Install at least one backend CLI and sign in once.
- Restart Claude Code, run
/mcp, and check thatagent-internis connected. - Ask for a tool by name, such as
opencode_status.
npm i -g @opencode/cli codex login npm i -g @github/copilot cursor-agent login
Add it to Claude Code from a terminal:
claude mcp add agent-intern -- uvx agent-intern
Or add this to .mcp.json in your project root (Claude Code) or .cursor/mcp.json (Cursor):
{
"mcpServers": {
"agent-intern": {
"command": "uvx",
"args": [
"agent-intern"
]
}
}
}Environment variables
AGENT_INTERN_NO_STATS: turns stats recording off.
Example prompts
- "Run opencode_status."
- "Ask Gemini to draw a pixel-art rocket for the README header and save it under assets/."
- "Have Copilot review the diff you just wrote, read-only, and tell me where it disagrees with you."
- "Summarise each of the six files in src/handlers in parallel with a swarm."
What to watch out for
- You need uv, and the opencode CLI installs through npm, so it needs Node.js with npm.
- You must install at least one backend CLI and sign in once.
- The opencode free models use a shared free tier, so they are rate-limited.
- The first start downloads the package, so it can take a little while.
- A plain
uvx agent-internpicks up a new release the next time your host starts the server; pin it withuvx agent-intern==X.Y.Zto decide when updates land. - Installing both the plugin and the MCP server gives Claude every tool twice.
- In a headless
agy -p, MCP tools are denied unless you allow them one rule per tool. - The Antigravity IDE is not verified yet.
- Grok, Kimi and Muse are experimental.
Not tested by tools4agents yet. Read the source repository before you install it.
Similar tools
Everything
The official test server. Exercises every part of the protocol, so it is the fastest way to check your setup works at all.
Filesystem
Gives an agent scoped read and write access to directories you name explicitly. The usual first MCP server people install.
Git
Read, search, and manipulate local Git repositories. Lets an agent inspect history and diffs without shelling out.
Sequential Thinking
Gives a model a scratchpad for multi step reasoning, letting it revise earlier steps instead of committing to a first answer.