Skip to main content
Glama
umum-ai

claude-codex-bridge

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault

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

CapabilityDetails
tools
{}
experimental
{
  "claude/channel": {}
}

Tools

Functions exposed to the LLM to take actions

NameDescription
codex_modelsA

List available Codex models. Pass nextCursor as cursor for another page.

codex_startA

Start a live Codex agent and return its threadId immediately after acceptance. Codex runs with full filesystem/network access and no approval prompts. Supply an absolute working directory and the complete task context; model is optional.

codex_messageA

Send direction to a running Codex turn, or start another turn in the same thread if it is idle. A race with turn completion can return an error: inspect status before retrying.

codex_stopA

Request interruption of a running Codex turn. The interrupted event confirms completion; this does not undo changes already made.

codex_resumeA

Attach a saved Codex thread after reconnecting. Use codex_message to continue its work.

codex_answerB

Answer a Codex question using its requestId and all question IDs. Ask the user when the answer requires their preference; otherwise answer from the task context.

codex_statusA

Read session state, pending questions and the latest answer. Omit threadId to list sessions. Does not wait for completion.

codex_eventsA

Read the bounded event journal after a sequence number. gap means older events were evicted. Use to recover results if channel delivery is unavailable; do not busy-poll.

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

TDQS

A4/5.0

Scored across 8 tools

Disambiguation5/5

Each tool maps to a distinct Codex lifecycle action: start, message, stop, resume, answer, status, events, and model listing. There is no meaningful overlap, and edge cases like codex_message versus codex_answer are clearly separated by their descriptions.

Naming Consistency4/5

All tools share a clean codex_ prefix and use snake_case consistently. Most names are imperative verbs, but models, status, and events are noun-style read operations, which is a minor deviation from a strict verb_noun pattern.

Tool Count5/5

Eight tools is a well-scoped size for a bridge server managing a live agent lifecycle. Each tool serves a necessary function without redundancy or bloat.

Completeness5/5

The set covers the full agent lifecycle: starting, sending messages, stopping, resuming, answering questions, checking status, and reading event history. There are no obvious gaps that would leave an agent unable to complete a workflow.

Maintenance

ActivityMaintained
ResponsivenessUnresponsive