AgentBridge
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
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 | {} |
| prompts | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| codex_runA | Delegate a task to the Codex CLI running as an external worker, using your existing Codex login (no API key). Model and reasoning effort are chosen per call and never inherited from global Codex config. Returns the worker's final result plus git-verified files_changed and scope_violations, and — unless skipped — a |
| antigravity_runA | Delegate a task to the Antigravity CLI ( |
| bridge_statusA | Report AgentBridge health: version, project directory, whether git-based scope verification will work here, whether the Codex and Antigravity CLIs are installed, their versions and sign-in state, per-model availability (available | unavailable | unverified) and the efforts each model accepts, resolved Antigravity model ids, active jobs, and warnings. Consumes no model inference — availability comes from each CLI's own catalogue, and is reported as |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| agentbridge | Size a goal, plan it, delegate it across the AI CLIs you have installed, verify the result and report. The main entry point. |
| agentbridge_setup | First run. Detect installed CLIs, prove each one works, interview the user, research their models, and write the routing profile. |
| agentbridge_status | Render bridge_status plainly: what is installed, signed in and available, and how the profile maps roles today. Costs no inference. |
| agentbridge_doctor | Diagnose the installation with real smoke jobs rather than version probes, and say exactly what to fix. |
| agentbridge_profile | Show the routing profile, explain why a role maps where it does, or re-derive a single role without redoing setup. |
| agentbridge_recon | Delegate one bounded repository investigation and report the answer. No implementation, no batches. |
| agentbridge_review | Review the working diff, or a named target, using a different model family than the one that wrote it. |
| agentbridge_resume | Read the run state and continue where a previous run stopped, without repeating finished work. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 3 tools
Each tool has a clearly distinct purpose: codex_run and antigravity_run target different external CLIs, and bridge_status is the health/status reporter. There is no meaningful overlap or ambiguity between them.
The two worker tools follow a consistent <cli>_run pattern, while bridge_status deviates by using a noun instead of a verb. This is a minor inconsistency in an otherwise predictable naming scheme.
Three tools is well-scoped for a bridge server: one for each external worker and one for status. Every tool earns its place and the count feels appropriate for the server's purpose.
The core workflows of delegating to either CLI and checking bridge health are covered. Minor gaps exist around job lifecycle management (e.g., listing or canceling active jobs), but these are not critical to the stated purpose.