Skip to main content
Glama
khoa-na

antigravity-mcp

by khoa-na
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Local

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
AGY_EXENoExplicit path to agy.exe. Auto-detected if not set.
AGY_MODELNoDefault model ID used by the agent.gemini-3.8-flash-high
CLOUD_CODE_URLNoSet only if using a custom local unlock proxy.
AGY_ACCESS_TOKENNoAccess token used for authentication on Linux and macOS.
AGY_AGENT_TIMEOUTNoTimeout in seconds before terminating long runs.900
AGY_DEFAULT_WORKSPACENoDefault workspace granted to the agent. Set to 'none' to isolate.process.cwd()

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

CapabilityDetails
tools
{
  "listChanged": false
}
prompts
{
  "listChanged": false
}
resources
{
  "subscribe": false,
  "listChanged": false
}
experimental
{}

Tools

Functions exposed to the LLM to take actions

NameDescription
ask-antigravityA

Ask the Google Antigravity AGENT (Gemini 3.8 Flash / Pro) — a coding agent WITH file tools (Read/Write/Bash), like a second Claude Code / Codex.

By default it works in the current directory (process workspace), so it can read and run your repo like Codex. Supports parallel multi-agent swarm execution and one-click session resume.

Parameters:

  • prompt: The task/code instruction.

  • resume: If True, automatically continues the most recent conversation session (--continue) without needing an ID.

  • prompt_file: Absolute path to a file containing the prompt (wins over prompt).

  • model: Model ID (default: gemini-3.8-flash-high, or gemini-3.1-pro-high, etc.).

  • system: Optional system instructions.

  • system_file: Absolute path to a file containing system instructions.

  • workspace: Directory granted to the agent via --add-dir (read+write). Default: current directory. Pass 'none' for isolated temp dir.

  • conversation_id: Resume/continue an existing session by conversation ID.

  • effort: Reasoning effort (low, medium, high).

  • cleanup: If true, delete prompt_file / system_file after execution.

  • save_artifact: If true, always saves response to an artifact file in .antigravity/artifacts/.

ask-geminiD

Alias for ask-antigravity.

review-diffA

Perform an automated, structured code review on current uncommitted or staged git changes. Runs in strictly read-only mode so it never modifies your repository files.

Parameters:

  • workspace: Path to the git repository (default: current workspace).

  • staged: If true, reviews only staged changes (git diff --staged). Otherwise reviews all uncommitted changes.

  • focus: Optional focus area (e.g. 'security', 'performance', 'logic bugs', 'code style').

  • model: Target model ID (default: gemini-3.7-flash-high).

check-quotaA

Retrieve and display live remaining quota percentages for all Google Antigravity models. Shows health status (Healthy / Low / Depleted) for each model.

generate-testsA

Generate thorough unit tests for a specified source code file.

Parameters:

  • target_file: Relative or absolute path to the file you want to test.

  • test_framework: Test framework (e.g. 'pytest', 'unittest', 'jest', 'go test'). Default: 'pytest'.

  • workspace: Workspace root directory.

  • write_to_file: If True, writes the generated tests directly into output_file.

  • output_file: Destination file path for generated tests (e.g. 'tests/test_auth.py').

  • model: Target model ID (default: gemini-3.8-flash-high).

pingB

Health check — verifies that the Antigravity CLI or backend responds.

list-modelsA

List available models in Google Antigravity.

Prompts

Interactive templates invoked by user choice

NameDescription
code-reviewPrompt template for structured, senior-level code review of a diff or implementation.
tdd-featurePrompt template for Test-Driven Development (TDD): write failing tests first, then implement.
security-auditSecurity audit prompt template focusing on OWASP Top 10, sanitization, authentication, and CVE patterns.

Resources

Contextual data attached and managed by the client

NameDescription
quota_resourceLive remaining quota percentages for Google Antigravity models.
models_resourceList of available models in Google Antigravity.
status_resourceActive Google Antigravity runtime status and configuration.
Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

B3.3/5.0

Scored across 7 tools

Disambiguation4/5

Most tools have clearly distinct purposes: asking the agent, reviewing diffs, generating tests, checking quota, listing models, and health checking. The only overlap is ask-gemini being explicitly documented as an alias for ask-antigravity, so while redundant, it does not create real selection ambiguity.

Naming Consistency5/5

All tools follow a consistent lowercase hyphenated verb-noun pattern: ask-antigravity, review-diff, check-quota, generate-tests, list-models. Even ping fits the simple verb style, and there are no mixed naming conventions.

Tool Count5/5

Seven tools is a well-scoped set for an Antigravity integration. Each tool covers a meaningful capability, and the only minor redundancy is the explicitly marked ask-gemini alias.

Completeness4/5

The server covers the main workflows: invoking the agent, reviewing changes, generating tests, checking quota/model availability, and health checks. Session resume and workspace control are handled through ask-antigravity parameters, so there are no critical gaps, though a dedicated session-management tool could be a potential enhancement.

Maintenance

ActivityMaintained
ResponsivenessNo issues

Resources