MCP Server

MCP server exposing Google Antigravity (agy CLI / Gemini): jobs, codebase analysis, file tools.

Local serverstdio

What is the MCP MCP server?

MCP server exposing Google Antigravity (agy CLI / Gemini): jobs, codebase analysis, file tools. That is what the mcp mcp server brings to an AI assistant: the same capability, reachable through the Model Context Protocol rather than a separate app or dashboard.

The short version

The idea: your primary model (Claude Desktop, Claude Code, Codex, …) stays lean and delegates the grunt work — web search, whole-repo analysis, scaffolding files, long-running edits — to a second agent running in the background. Jobs are async and unlimited: you fire a task, get a jobId instantly, and poll for the result while the caller keeps working.

The tools it exposes

The server publishes 14 tools. What each one is for:

  • use_antigravity — Delegate a task (async). Returns a jobId immediately. Params: prompt, thinking_depth (low/high), add_dirs (folders agy may read/write), auto_approve
  • antigravity_continue — Continue the previous conversation (or a specific conversation_id) with a new prompt
  • antigravity_result — Get a job's result by jobId. Supports long-polling via optional wait_ms and poll_interval_ms. While running, returns a lean payload ({ status, jobId
  • antigravity_jobs — List recent jobs and statuses
  • antigravity_cancel — Kill a running job by jobId
  • antigravity_cleanup — Delete old job files by age (older_than_hours)
  • antigravity_create_folder — Create a folder (recursive), done directly by the server
  • antigravity_create_file — Create/write a text file (creates parent dirs); overwrite:false to protect existing
  • antigravity_create_tree — Build a whole folder/file tree from a JSON spec
  • antigravity_read_file — Read a UTF-8 text file (instant); optional max_bytes cap
  • antigravity_list_dir — List a folder's entries
  • antigravity_models — List available models (agy models)

What it needs from you

Configuration is passed through the environment: AGY_MODEL, AGY_AUTO_APPROVE. Treat anything key-shaped as a real credential — scope it to the minimum the server needs, and rotate it if it ever lands in a shared config.

| Need | Why | |------|-----| | Node.js 18+ | Runs this server (ships with npx/npm). Check: node --version | | Antigravity CLI (agy) | The agent this server drives. Install below. | | An MCP client | Claude Desktop, Claude Code, or anything that speaks MCP | ---

Getting it running

mcp-server-google-antigravity on npm is all you need. Most clients run it directly, so configuration is a few lines and a restart.

How it compares

This sits in the planning and project tracking group, where several servers overlap in what they claim to do but differ sharply once you actually set them up. MCP's toolset — use_antigravity, antigravity_continue, antigravity_result and 11 more — is a fair guide to whether it matches your workflow. It is maintained by TurkerYakup; worth a glance at recent repository activity before you build anything load-bearing on it.

We check each listing at SyncDev against the project's documentation before it goes live — if something here drifts out of date, it is a bug worth reporting.

Things to watch

  • 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 14 tools registered it takes up a noticeable share of the context window; turn it off in projects that never touch MCP.
  • Keep per-call confirmation enabled while you learn its behaviour; it is the cheapest safeguard you have.

Available tools

ToolWhat it does
use_antigravityDelegate a task (async). Returns a jobId immediately. Params: prompt, thinking_depth (low/high), add_dirs (folders agy may read/write), auto_approve, new_project, model, mode (plan/accept-edits), agent, project, sandbox,
antigravity_continueContinue the previous conversation (or a specific conversation_id) with a new prompt.
antigravity_resultGet a job's result by jobId. Supports long-polling via optional wait_ms and poll_interval_ms. While running, returns a lean payload ({ status, jobId, chars, tail, startedAt }). On done/error, returns full record (includi
antigravity_jobsList recent jobs and statuses.
antigravity_cancelKill a running job by jobId.
antigravity_cleanupDelete old job files by age (older_than_hours).
antigravity_create_folderCreate a folder (recursive), done directly by the server.
antigravity_create_fileCreate/write a text file (creates parent dirs); overwrite:false to protect existing.
antigravity_create_treeBuild a whole folder/file tree from a JSON spec.
antigravity_read_fileRead a UTF-8 text file (instant); optional max_bytes cap.
antigravity_list_dirList a folder's entries.
antigravity_modelsList available models (agy models).
antigravity_agentsList available agent profiles (agy agents).
antigravity_healthCheck that agy is reachable + report effective defaults. **Run this first when something's off.**

How to install the MCP MCP server

{
  "mcpServers": {
    "google-antigravity": {
      "command": "npx",
      "args": ["-y", "mcp-server-google-antigravity"],
      "env": {
        "AGY_MODEL": "your-value",
        "AGY_AUTO_APPROVE": "your-value"
      }
    }
  }
}

Add to claude_desktop_config.json, then restart Claude Desktop.

Configuration

| Need | Why | |------|-----| | Node.js 18+ | Runs this server (ships with npx/npm). Check: node --version | | Antigravity CLI (agy) | The agent this server drives. Install below. | | An MCP client | Claude Desktop, Claude Code, or anything that speaks MCP | ---

VariableDescriptionRequired
AGY_MODELConfiguration value read at startup.Optional
AGY_AUTO_APPROVEConfiguration value read at startup.Optional

Example prompts to try

  • Use MCP to use antigravity.
  • Use MCP to antigravity continue.
  • Use MCP to antigravity result.

Frequently asked questions

It connects MCP to MCP-compatible AI assistants such as Claude and Cursor, exposing 14 tools (use_antigravity, antigravity_continue, antigravity_result, and more) that the assistant can call on your behalf. Instead of copying data back and forth by hand, the assistant works with MCP directly.