</>MCP Agents Market
Agent

claude_codex_bridge

by SeemSeam3.5kPythonUpdated 2026-09-01

Visible multi-agent CLI workspace for mixing Codex, Claude, Gemini, Kimi, Qwen, Cursor, Copilot, Pi, OpenCode, and other AI coding agents

CodexClaudeGeminiCursorGitHub Copilot

claude_codex_bridge (CCB) is a terminal-based multi-agent workspace that orchestrates collaboration between 16+ AI coding CLI families including Codex, Claude, Gemini, Kimi, Qwen, Cursor, Copilot, Pi, and OpenCode. It provides visible pane management, inter-agent communication protocols, shared project memory, and workflow delegation through a terminal UI with tmux-based layout control. Developers use CCB to coordinate complex multi-agent tasks like code review pipelines (A → B → C) or parallel analysis workflows (A,B → C), with persistent project state, mobile remote control, and optional rich media terminal support through WezTerm.

Key Features

Multi-agent orchestration supporting 16 CLI provider families with visible terminal panes and direct takeover
Stable inter-agent communication for complex collaboration graphs (sequential, parallel, branching workflows)
Background daemon preserving project state, queues, and conversation history across UI sessions
Android mobile app with voice control, file transfer, terminal access, and Relay/Tailscale/LAN connectivity
Agent Roles Spec support for packaging specialist agents with skills, memory, and dependencies into installable Role Packs
Rich media terminal mode with WezTerm integration for file browsing, editing, and media preview
Isolated per-agent provider authentication, API configuration, and workspace inheritance without modifying global state
Web-based configuration UI for visual topology editing, model/thinking controls, and agent memory management

Use Cases

  • 01Multi-stage code review pipelines where one agent writes code, another reviews architecture, and a third validates tests
  • 02Parallel research and implementation workflows coordinating multiple specialized agents on different aspects of a feature
  • 03Remote mobile control of local AI coding workflows with voice input and file synchronization
  • 04Cross-provider model comparison by running identical tasks through Claude, Codex, and Gemini simultaneously
  • 05Team-based agent collaboration using shared project memory for handoff conventions and long-lived context
  • 06Workflow orchestration with specialist Role Pack agents for frontend, mobile, architecture review, or domain-specific tasks

Related Agents

View more

claude_codex_bridge — FAQ

What is claude_codex_bridge?+

CCB is a terminal-based multi-agent workspace that coordinates collaboration between 16+ AI coding CLI tools like Codex, Claude, Gemini, Kimi, and Cursor in visible, controllable workflows. It manages inter-agent communication, shared memory, and persistent project state through a daemon-backed TUI.

How do I install CCB?+

Install via npm with 'npm install -g @seemseam/ccb@latest', or download the release tarball and run './install.sh install'. For Windows x64 beta, download the ZIP from GitHub releases and run the PowerShell installer. After installation, run 'ccb' from your project directory to launch the workspace.

Which AI clients does CCB work with?+

CCB integrates with Codex, Claude, Gemini, Kimi, Qwen, Cursor, GitHub Copilot, Pi, OpenCode, Grok, and several other CLI-based AI coding tools. It manages these as visible terminal panes within a coordinated multi-agent workspace rather than acting as a plugin for IDEs.

Do I need API keys for CCB?+

You need valid authentication for whichever AI provider CLIs you want to use (Codex, Claude, Gemini, etc.). CCB inherits credentials from each provider's native authentication and projects them into isolated per-agent homes. The CCB orchestration layer itself does not require separate API keys.

Is CCB free to use?+

CCB itself is open-source and free. However, you pay for the underlying AI provider services (OpenAI, Anthropic, Google, etc.) according to their pricing. CCB adds no additional usage fees beyond the cost of the AI models you invoke.

What platforms does CCB support?+

CCB runs on Linux, macOS, and WSL, with native Windows x64 in beta. It requires Python 3.10+, tmux (or Herdr on Windows), and optionally WezTerm for rich mode. The Android mobile app supports remote control over LAN, Tailscale, or encrypted Relay.

How do I install claude_codex_bridge?+

Open the source repository on GitHub and follow its README. claude_codex_bridge is a agent — MCP Agents Market links you directly to the official repo.

Is claude_codex_bridge free?+

claude_codex_bridge is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.

Related searches