Skip to main content
Glama

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
AGY_BINNoOptional path to the AGY CLI executable. If not set, agy-bridge looks for it at ~/.gemini/bin/agy or on PATH.

Instructions

Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.

Run Google Antigravity (AGY) reviewer agents and converse with them. Start with agy_start, continue with agy_say, and use agy_wait if a turn is still running. Reviews take minutes. AGY reads the project and may run tests, but must not edit files under review; notes.tree_changed reports any change it made.

Capabilities

Features and capabilities supported by this server

Protocol revision2025-11-25

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

Tools

Functions exposed to the LLM to take actions

NameDescription
agy_list_agentsA

List AGY agents available to agy_start and the model slugs AGY offers.

Model slugs may carry an effort level (e.g. "-high"); pass either a full slug, or a base
slug plus `effort`. Never assume a fixed list: always read it here.
agy_startA

Open a new AGY review session and send the first message.

agent: an AGY agent, e.g. code-reviewer; agy_list_agents lists those installed.
cwd: absolute path of the project to review (AGY runs there).
message: what to review and why: diff or file/doc paths, intent, how to run tests.
model / effort: optional; omit to use AGY's default. See agy_list_agents.
schema: optional JSON schema for the answer; defaults to the review schema
  (verdict, summary, findings, questions_for_claude).
wait_seconds: how long to wait for the answer before returning status RUNNING
  (the turn keeps going; collect it with agy_wait).
Returns the session_id, the review, and notes (denied actions, files AGY changed).
agy_sayA

Send the next message in an AGY session: fixes made (with the new diff), answers to AGY's questions, or arguments against a finding. Resumes the conversation automatically if its process has ended (e.g. in a later Claude session). model/effort default to the session's previous choice, except that a new model starts with no effort unless one is given. Changing them restarts the process on the same conversation.

agy_waitA

Keep waiting for a turn that agy_start/agy_say returned as RUNNING, and return its result when done. Returns the last result again if no turn is running.

agy_statusB

What an AGY session is doing right now: idle or working, current turn, elapsed time, and the last tool steps AGY took. Use it when the user asks for progress.

agy_resumeA

Re-open a stored AGY session in a fresh process, by session_id or conversation_id. For a conversation the bridge has never seen, also pass agent and cwd. agy_say also resumes automatically, so this is only needed to warm a session up front.

agy_sessionsB

List stored AGY sessions, most recently used first. Filter by project with cwd.

agy_endA

Stop a session's AGY process. The session stays resumable unless forget=True, which also deletes it from the registry and removes its scratch, backup and log files.

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

TDQS

A3.7/5.0

Scored across 8 tools

Disambiguation4/5

Each tool maps to a distinct lifecycle step (start/say/wait/status/resume/sessions/end), and descriptions clearly delimit agy_status (quick peek) from agy_wait (blocking) and agy_resume (warm-up only). The one real overlap is agy_resume vs agy_say's automatic resume, which the description explicitly acknowledges.

Naming Consistency4/5

All tools share the agy_ prefix and snake_case, with mostly verb-led names (agy_start, agy_say, agy_list_agents). agy_status and agy_sessions are noun-only, a minor deviation but still predictable and readable.

Tool Count5/5

Eight tools is well-scoped for a session-bridge server, each corresponding to a genuine lifecycle operation. Nothing feels padded, and no obvious capability is crammed into a generic catch-all tool.

Completeness4/5

Full lifecycle coverage: agent discovery, start, converse, wait, resume, list, status, and end with optional forget. The main gap is no way to retrieve a past turn's transcript or full history from an ended session, though status and prior results partially compensate.

Maintenance

ActivityMaintained
ResponsivenessNo recent community issues