Antigravity MCP Bridge
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| NGROK_AUTHTOKEN | Yes | Your ngrok authtoken, required to create a secure public tunnel to the local MCP server. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| run_system_commandC | Executes a shell/PowerShell command on the local system (e.g. python, npm, git, tests, pip). Includes safety filtering and configurable timeout (default: 180s). |
| read_fileC | Reads the content of a file from the local filesystem. |
| write_fileB | Creates or overwrites a file on the local filesystem with specified content. Automatically creates parent directories if they don't exist. |
| edit_fileC | Performs a precise surgical search-and-replace edit in an existing file. Avoids having to rewrite the entire file when making targeted code changes. |
| append_fileB | Appends text to the end of an existing file (or creates it if it doesn't exist). |
| batch_write_filesA | Creates or updates multiple files in a single tool call. Reduces permission prompts from N to 1. Parameters: files: Dictionary of {"relative/or/abs/filepath": "file_content"} base_dir: Optional root directory (defaults to server CWD) |
| run_batch_commandsA | Executes multiple shell/PowerShell commands sequentially in a single tool call. Reduces permission prompts from N to 1. Parameters: commands: List of shell command strings to execute in order. working_dir: Working directory for all commands. stop_on_error: If True, halts execution if any command returns non-zero exit code. timeout_per_command: Max seconds per command (default: 180s). |
| create_full_projectA | Creates an entire project directory, writes all code files, and runs initial setup/test commands in a SINGLE tool call with 1 permission confirmation. Parameters: project_name: Name of the project folder to create. files: Dictionary of {"rel/path/filename.ext": "file content"} setup_commands: Optional list of commands to run inside the new project (e.g. ["pip install -r requirements.txt", "python test.py"]) working_dir: Parent directory where project folder will be created (defaults to CWD). |
| list_directoryC | Lists files and directories at the specified path with file sizes. |
| git_quick_statusB | Returns high-level Git status: active branch, changed files, untracked files, and recent commit. |
| run_agent_taskC | Launches an autonomous Antigravity AI agent task using ANTIGRAVITY's credits and models. Routes the task into a real Antigravity conversation via message injection. |
| get_agent_statusB | Fetches the live status and output of a running or completed agent task. |
| terminate_taskA | Terminates or cancels a running agent task by task ID. |
| get_bridge_historyB | Returns the full shared history of all tool calls made through this bridge. |
| save_session_noteB | Saves a note or memory to the shared bridge session log. |
| get_session_notesB | Retrieves all saved session notes from the shared bridge log. |
| sync_project_to_geminiB | Called by Antigravity to push a complete project status report and architectural summary into the shared bridge memory. Gemini Spark can read this anytime to understand your exact project state. |
| request_spark_connected_app_actionB | Dispatches a task from Antigravity to Gemini Spark requesting execution via Spark's connected apps (e.g. @Canva, @Google Drive, @Google Docs, @Google Keep, @YouTube, @Gmail, @Dropbox). Spark will read this request on sync and execute the tool action in the Google ecosystem. |
| get_spark_connected_apps_catalogA | Returns the complete list and capabilities of external tools & Google Workspace apps connected to Gemini Spark that Antigravity can orchestrate. |
| list_antigravity_conversationsA | Lists ALL Antigravity projects/conversations with their real names, conversation IDs, last active time, message count, artifact count, and task count. |
| inject_messageC | Injects a message directly into any Antigravity conversation's inbox. |
| send_spark_to_antigravity_taskA | Sends a crystal-clear, structured task brief from Gemini Spark to Antigravity IDE. Automatically generates a formatted markdown instruction envelope with Task ID, objectives, context, step-by-step actions, and reporting instructions. Parameters: objective: Clear 1-2 sentence primary goal. context: Optional background, architectural details, or file paths. required_actions: Optional ordered list of specific steps (e.g. ["write tests", "run pytest", "fix bugs"]). conversation_id: Target Antigravity conversation UUID (if None, targets most recent active). working_dir: Target working folder on disk. |
| get_antigravity_agent_reportB | Retrieves the structured status report and latest responses from Antigravity. Shows completion notes, modified files, test outputs, and executive summary. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 23 tools
Multiple tools have overlapping purposes, such as run_system_command vs run_batch_commands, write_file vs batch_write_files, and run_agent_task vs send_spark_to_antigravity_task vs inject_message. The boundaries between session notes, bridge history, and agent status reports are also unclear, creating significant potential for misselection.
Most tool names follow a verb_noun pattern, but the verbs are inconsistent across similar actions (e.g., run_agent_task, send_spark_to_antigravity_task, inject_message) and some names like git_quick_status do not follow the pattern. The naming is readable but not systematically consistent.
With 23 tools, the set is at the high end of the 'heavy but not extreme' range. The broad scope of filesystem, shell, agent, notes, and external app operations partially justifies the count, but it feels larger than necessary and could be consolidated.
The toolset covers core filesystem operations, command execution, batch operations, agent task management, session notes, and external app integration. Minor gaps exist such as no explicit delete file/note tool and no direct list-agent-tasks function, but the primary workflows are well represented.