Skip to main content
Glama
rauls-kjarners

claude-to-agy

by rauls-kjarners
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Local

claude-to-agy

CI CodeQL codecov PyPI version Static Badge

A lightweight MCP bridge that lets Claude Code delegate heavy tasks to the Antigravity CLI (agy) - saving context window and tokens for what matters.

What It Does

Registers a delegate_to_agy MCP tool that Claude automatically uses when it encounters:

  • Large files (>200 lines) - logs, dumps, generated code

  • Multi-file analysis (>3 files at once)

  • Deep searches - git log, git diff, grep

  • Web lookups - documentation, external knowledge

  • Adversarial review / plan critique - always delegated

Claude sends a prompt + file paths → the bridge runs agy CLI → returns the result.

Related MCP server: agy-bridge

Requirements

Installation

Note: The command below uses uvx to run the server without manual installation. If you don't have uv installed, you can substitute uvx with npx -y or pipx run.

# 1. Register the MCP server globally via uvx
claude mcp add -s user claude-to-agy uvx claude-to-agy

# 2. Download the delegation rules into your current project
curl -o CLAUDE.md https://raw.githubusercontent.com/rauls-kjarners/claude-to-agy/main/CLAUDE.md

That's it. Claude will now automatically delegate heavy tasks to Antigravity CLI in any project that has the CLAUDE.md file.

Tip: To enable globally without copying CLAUDE.md per project, add the rules to ~/.claude/CLAUDE.md instead.

Using as a Skill

This project also includes a SKILL.md file, which is the standard format for reusable Claude Code skills. If your setup supports skills, you can use it instead of manually copying CLAUDE.md:

claude skill add https://raw.githubusercontent.com/rauls-kjarners/claude-to-agy/main/SKILL.md

Note: You still need the MCP server registered (step 1 above). The skill provides the rules, MCP provides the tool.

Configuration

All settings are optional environment variables:

Variable

Default

Description

AGY_CONNECT_TIMEOUT

60

Seconds to start the agy process

AGY_TOTAL_TIMEOUT

1200

Hard timeout for entire execution

How It Works

User → Claude Code → MCP bridge (FastMCP) → agy CLI → Gemini API
                   ←                      ←         ←

  1. CLAUDE.md instructs Claude when to delegate

  2. Claude calls delegate_to_agy(prompt, cwd, files?) via MCP

  3. bridge.py prepends file paths to the prompt

  4. Runs agy --dangerously-skip-permissions --add-dir <cwd> -p "<prompt>"

  5. Returns the text output cleanly or raises an exception natively handled by FastMCP

Development

# Linting & Formatting
uv run ruff check .
uv run ruff format .

# Type Checking
uv run pyright

# Tests
uv run pytest

# Pre-commit Hooks (Run before committing)
uv run pre-commit install

License

MIT

Available Tools

1 tool
delegate_to_agyB

Delegate complex reasoning or deep search to Antigravity CLI.

ParametersJSON Schema
NameRequiredDescriptionDefault
promptYesThe detailed prompt/instructions for Antigravity CLI.
cwdYesThe working directory of the project (absolute path).
filesNoOptional list of absolute file paths to include as context.

Output Schema

ParametersJSON Schema
NameRequiredDescription
resultYes

TDQS

B3.1/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations provided, the description carries full burden for behavioral disclosure. It fails to mention whether the tool is read-only, modifies state, is asynchronous, or has error/response characteristics beyond the output schema existence.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is a single, concise sentence. It is appropriately short for the complexity but could benefit from a bit more structure (e.g., separating purpose and usage). Still, it is not verbose.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness2/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Despite having an output schema and fully described parameters, the description omits critical context such as how the delegated task behaves, expected response format, or any side effects. An agent would lack sufficient information to use the tool confidently.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so each parameter is already described in the schema. The tool description does not add additional semantic meaning beyond restating 'complex reasoning or deep search', which provides context but not parameter-specific detail.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description uses a specific verb ('delegate') and identifies the resource ('Antigravity CLI') and its purpose ('complex reasoning or deep search'). It clearly communicates what the tool does without ambiguity.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

No guidance is provided on when to use this tool versus alternatives, or when not to use it. Since there are no sibling tools, the absence is less critical, but the description does not help the agent decide contextually.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

  1. 1 tool updatev0.1.0
    • First observeddelegate_to_agy
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

B3.4/5.0

Scored across 1 tool

Disambiguation5/5

With only one tool, there is no risk of confusion between tools. The single tool's purpose is clearly described, so an agent can easily identify what it does.

Naming Consistency5/5

The tool name follows a clear verb_noun pattern ('delegate_to_agy'), consistent with common conventions. With only one tool, naming consistency is inherent.

Tool Count3/5

One tool is sparse for a typical MCP server. While the server serves a narrow purpose (delegation to AGY), the single tool may feel insufficient for complex workflows, but it is not excessively heavy.

Completeness3/5

The server covers its stated purpose of delegation, but lacks additional tools for managing delegations (e.g., status checks, result retrieval, error handling). There are minor gaps that could affect agent reliability.

Maintenance

ActivityInactive
ResponsivenessUnresponsive

Resources

Related MCP Connectors

Related MCP Servers

  • agy-bridge

    sshahzaiib
    A
    license
    A
    quality
    B
    maintenance
    MCP bridge that lets Claude Code delegate heavy tasks to the Antigravity CLI (agy) — purpose-built tools, model routing with fallback, session continuity, and output truncation to save Claude's context and tokens.
    6
    111 npm
    51
    MIT

  • agent-intern

    SinanTufekci
    A
    license
    A
    quality
    A
    maintenance
    An MCP server that bridges Claude Code with Antigravity CLI using a Swarm Agent architecture to optimize local development workflows and minimize LLM token costs. Includes a web UI for monitoring agent workflows.
    24
    451 PyPI
    22
    MIT

  • antigravity-mcp

    k12club
    A
    license
    Not graded
    quality
    C
    maintenance
    An MCP server that lets Claude Code call the Google Antigravity CLI (agy) headlessly for a second opinion from a different model family, or to have agy read project files on Claude's behalf so large files never enter Claude's context window.
    18 npm
    MIT