agy-headless-bridge MCP server
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| AGY_PATH | No | Absolute path to the agy binary. If not set, auto-detected. | |
| AGY_BRIDGE_TIMEOUT | No | Seconds before a call is killed. | 180 |
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 | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| agy_askB | Send a one-shot prompt to the Google Antigravity CLI (agy) and return its response. Use to delegate a focused coding, debugging, or reasoning task to Gemini via Antigravity. |
| agy_researchB | Ask Antigravity to research a topic deeply. |
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 2 tools
agy_ask and agy_research both send a prompt to the same Antigravity CLI and return a response, so the boundary is blurred: a user could fulfill research needs by calling agy_ask with a 'research this' prompt. The description hints at a depth/intent difference, but nothing structural distinguishes the two tools (no separate parameters, output shapes, or guarantees).
Both tools use the same agy_ prefix with a single lowercase snake_case verb (ask, research), forming a clean, predictable pattern. There is no mixing of conventions or verb styles.
Two tools is thin for a bridge server, and one of them (agy_research) appears to be a thin wrapper around the other, so the surface feels under-scoped rather than deliberately minimal. It is not extreme, but there is little room for distinct capability.
The server covers one-shot prompt delegation but lacks any session/multi-turn continuation, file or workspace context parameters, or status/cancel operations that a CLI bridge typically needs. Core workflow (ask something and get an answer) works, but agents will hit dead ends for stateful or long-running tasks.