[KongchangAI]
Product

Codex Claude Bridge

An open-source multi-AI collaboration bridge tool that uses a local MCP server to let OpenAI Codex act as a planner orchestrating Claude Code to execute specific coding tasks. It supports Git worktree isolation, session management, and result review, enabling programmatic orchestration between two AI programming assistants.

Timeline (last 90 days)

Oct 12

Codex Claude Bridge is an open-source project that lets OpenAI's Codex act as an orchestrator, delegating coding tasks to Claude Code for execution and reviewing changes, with the entire workflow connected through a local MCP bridge layer.

Unverified50%
Oct 12

Developers only need to give Codex three things: a goal, a workspace, and acceptance criteria, with the remaining coordination handled automatically by the bridge layer.

Unverified50%
Oct 12

Codex Claude Bridge supports session management, allowing the creation of new Claude Code sessions or reconnection to existing sessions via UUID to ensure context continuity.

Unverified50%
Oct 12

Codex Claude Bridge uses isolated delegation, assigning tasks in separate Git worktrees and reviewing and merging afterward, so that experimental changes do not contaminate the main branch.

Unverified50%
Oct 12

Codex Claude Bridge allows independent adjustment of interaction frequency and reasoning effort, and can reuse previous inspection results when the source snapshot, environment, and scope match.

Unverified50%
Oct 12

Codex Claude Bridge preserves existing Claude Code model configurations, including third-party providers.

Unverified50%
Oct 12

Codex Claude Bridge supports both Windows and macOS platforms; Orca is optional and a regular terminal works fine.

Unverified50%
Oct 12

Running Codex Claude Bridge requires Codex CLI, a working Claude Code configuration, Node.js 20+, and Git.

Unverified50%
Oct 12

Codex Claude Bridge is installed via Codex's plugin marketplace using two commands: codex plugin marketplace add and codex plugin add.

Unverified50%
Oct 12

The author of Codex Claude Bridge is seeking feedback on three aspects: session recovery, task handoff, and coordination overhead.

Unverified50%

1 more timeline events

All Facts (11)

Unverified

Codex Claude Bridge is an open-source project that lets OpenAI's Codex act as an orchestrator, delegating coding tasks to Claude Code for execution and reviewing changes, with the entire workflow connected through a local MCP bridge layer.

50%
Unverified

Developers only need to give Codex three things: a goal, a workspace, and acceptance criteria, with the remaining coordination handled automatically by the bridge layer.

50%
Unverified

Codex Claude Bridge supports session management, allowing the creation of new Claude Code sessions or reconnection to existing sessions via UUID to ensure context continuity.

50%
Unverified

Codex Claude Bridge uses isolated delegation, assigning tasks in separate Git worktrees and reviewing and merging afterward, so that experimental changes do not contaminate the main branch.

50%
Unverified

Codex Claude Bridge allows independent adjustment of interaction frequency and reasoning effort, and can reuse previous inspection results when the source snapshot, environment, and scope match.

50%
Unverified

Codex Claude Bridge preserves existing Claude Code model configurations, including third-party providers.

50%
Unverified

Codex Claude Bridge supports both Windows and macOS platforms; Orca is optional and a regular terminal works fine.

50%
Unverified

Running Codex Claude Bridge requires Codex CLI, a working Claude Code configuration, Node.js 20+, and Git.

50%
Unverified

Codex Claude Bridge is installed via Codex's plugin marketplace using two commands: codex plugin marketplace add and codex plugin add.

50%
Unverified

The author of Codex Claude Bridge is seeking feedback on three aspects: session recovery, task handoff, and coordination overhead.

50%
Unverified

Orchestration overhead introduces additional latency and token consumption with each round-trip between Codex and Claude Code, and when tasks are broken down too granularly, the coordination cost may exceed the cost of direct execution.

50%

Source Articles