CodexBridge acts as a sophisticated gateway, connecting chat platforms, primarily WeChat, to a central Codex AI engine. It enables users to interact with advanced AI capabilities, manage AI-driven workflows, and maintain conversational context directly from their chat interfaces. The project targets power users and developers seeking to integrate AI assistance seamlessly into their daily communication and task management, offering benefits like enhanced productivity through chat-based automation and intelligent agent interaction.

How It Works

The architecture employs a modular design where platforms function as adapters to a unified Codex execution engine. This approach centralizes AI processing within Codex, ensuring consistency and leveraging Codex's thread state as the definitive source of truth. It supports dynamic switching between different backend provider profiles within Codex and facilitates rich interaction through a comprehensive set of text-based slash commands, designed for natural chat integration.

Quick Start & Requirements

• Primary Install: npm install
• Prerequisites: Node.js >= 24, npm, a working Codex CLI login (npm install -g @openai/codex@latest --include=optional).
• Setup: Run npm run typecheck and npm test for validation. WeChat login is initiated with npm run weixin:login, followed by npm run weixin:serve to start the bridge loop.
• Deployment: Scripts are provided for setting up Linux (systemd), Windows (Task Scheduler), and macOS (launchd) as background services for unattended operation.
• Docs: Full command reference available at docs/usage/weixin-slash-commands.md.

Highlighted Details

• Extensive WeChat slash command set for managing AI sessions, including thread manipulation (/threads, /open), skill and plugin discovery (/skills, /plugins), automation scheduling (/auto), personal assistant functions (/as, /log, /todo), and provider/model configuration (/provider, /model).
• OpenAI-compatible Responses adapter supports numerous non-OpenAI providers (DeepSeek, MiniMax, Qwen, OpenRouter, Kimi, Gemini, iFlow) via environment variable configuration, standardizing their integration.
• Media tooling leverages project-managed ffmpeg/ffprobe binaries for portable image normalization and video thumbnail generation across platforms.
• Personal assistant features (/as) allow natural language commands for creating, updating, and managing logs, todos, and reminders, with Codex determining the action type.

Maintenance & Community

The project's current focus is on stabilizing the WeChat + Codex integration, with specific workstreams (codex-gateway, mission-control, codex-native-api) noted as paused. No specific community links or contributor information were detailed in the provided text.

Licensing & Compatibility

License information is not explicitly stated in the provided README content.

Limitations & Caveats

Several components, including the codex-gateway and mission-control packages, are currently paused. The WebSocket runtime for the local OpenAI-compatible adapter is disabled pending the development of an upgrade handler. The absence of explicit licensing information may pose an adoption risk for certain use cases.