gitmyhub

feishu-codex-bridge

TypeScript ★ 217 updated 26d ago

Run local Codex CLI from Feishu/Lark chat, with sessions, attachments, and background service support.

A bridge that connects the Feishu or Lark chat apps to OpenAI Codex so you can type coding instructions in chat and have Codex run terminal commands and edit files on your local machine in real time.

TypeScriptNode.jsOpenAI CodexFeishu APILark CLIsetup: moderatecomplexity 3/5

This tool connects the Feishu and Lark chat apps to OpenAI's Codex coding assistant, letting you send messages in chat and have Codex run commands and edit files on your own computer in response. Instead of switching to a terminal window, you type instructions to a bot in a Feishu or Lark conversation, and the bot runs Codex locally and streams the output back to the chat.

The first-run setup is designed to minimize manual installation. The bridge checks whether Codex and a companion tool called Lark CLI are installed, and if either is missing, it installs them into a private folder in your home directory rather than a system-wide location. This avoids permission problems on managed company machines. You scan a QR code to link to a Feishu or Lark account and create a bot app through a guided wizard.

Once running, each chat conversation or topic group gets its own Codex session. You can switch which folder on your computer the bot works in, send files and images as attachments that Codex receives locally, and resume previous sessions if a conversation is interrupted. Slash commands control the main behaviors: /new resets the current session, /cd changes the working folder, /config adjusts concurrency, tool visibility, and reasoning settings, and /stop halts a running task.

On macOS, the bridge can be installed as a background service using the system's built-in launchd mechanism, so it keeps running after you close the terminal. Logs are stored in a standard folder under your home directory.

Access control settings let you restrict which users or chat groups can interact with the bot, which is useful when you want to share the assistant with a small team while keeping control over what runs on your machine.

Where it fits