wow-ai CLI
A bridge between World of Warcraft and local coding agents: type a task in the in-game chat and get the answer back in the game.
- 182 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- The bridge runs command-line agents in your project folder and they can edit files and run commands.
- The bridge captures the game screen four times a second and creates about 15,000 files in the client folder.
Install
- Works in
Manual install
node setup.js --project <папка проекта>Run inside a clone of `git clone https://github.com/chelinho139/wow-ai`; then restart WoW and `npm start`.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The bridge runs command-line agents in your project folder and they can edit files and run commands.
- The bridge captures the game screen four times a second and creates about 15,000 files in the client folder.
This is third-party code. Review the repository files before installing.
What it does
wow-ai links the World of Warcraft: Forever addon with coding agents on your machine: Claude Code, Codex, Grok Build, Antigravity CLI and Hermes Agent. You type a task in the in-game chat, the bridge runs the chosen agent headless in the right folder and the answer returns to the chat and a bubble window. Each chat keeps its own persistent agent session and they run in parallel. Progress is visible: action count, elapsed time, the files the agent edits and the commands it runs. The agent knows your character, level, zone, talents, professions and quest log, and draws routes and marks on the world map. Ready-made macros come with a button in the reply. Transport uses no code injection and no game memory reads: the addon draws the message as colored squares in a screen corner, and the bridge captures that corner four times a second and decodes it. It runs on Windows, Linux under Wine and macOS.
Who it is for. WoW players who write code or run projects with an agent and want to hand it tasks without leaving the game.
Good fit when
- A long dungeon queue is on and an agent needs to run the project tests.
- The agent should draw an ore or herb gathering route on the map based on your skill.
- A class macro is wanted and the agent delivers it with a Create macro button.
- Several chats must work with different agents and folders at once.
Not a fit when
- You do not play World of Warcraft: Forever: the bridge is built for it.
- You need headless server work with no screen: the bridge captures the game screen.
- Running agents with access to work projects from a game process is forbidden.
Example request
Ask the agent in the game chat why a test fails in the project and show the list of edited files in the game.Limitations
At least one command-line agent must be installed and logged in: `claude`, `codex`, `grok`, `agy` or `hermes`. Installation copies the addon into the client folder and creates about 15,000 tiny files for the channel. On Linux the game runs under Wine on an X11 session with python3, and macOS needs a native client. The agent works in the chosen project folder with the rights of your account, and the Allow and retry button extends the list of permitted commands. The project grew out of wow-claude, and rerunning `node setup.js` carries old chats and settings over.
How to disable. Stop the bridge (Ctrl+C), disable the WoW AI addon on the AddOns screen and delete the client `Interface\AddOns\WoWAI` folder and the repository clone.
FAQ
Which agents are supported?
Claude Code, Codex, Grok Build, Antigravity CLI (`agy`) and Hermes Agent. The default agent is the `agent` field in `bridge/config.json`, and `/wow-ai agent codex` switches the current chat.
How does the game link work?
The addon draws the message as colored squares in the top-left corner, the bridge captures that corner and runs the agent, and the reply is written into a pool of pre-made addons the game loads on a timer.
Does this break game rules?
The project states that no code is injected, no game memory is read and no input is generated. The addon uses documented APIs only, and the bridge reads the screen and writes ordinary files.
What if the agent needs a command outside the allowlist?
An Allow and retry button appears in the reply for Claude and Grok. It permits the command and retries the request.
How do I show ore and herbs on the map?
With `/wow-ai map ore` and `/wow-ai map herb`: spawns are filtered by your gathering skill.
Discussion
No questions or reviews yet.
Related
CLI toolsA realtime voice layer on top of Claude Code, Codex and OpenCode that keeps talking and listening while background tasks run in parallel
The /myoffice command opens a browser office showing what every agent on your machine is doing: thinking, waiting for you or done.
Agent Client for Obsidian
Agent Client
An Obsidian plugin that runs Claude Code, Codex and other agents via the Agent Client Protocol right inside your notes vault
Sound and on-screen notifications for coding agents: game voice lines when the agent finishes or waits for permission
Voice control of Claude Code and Codex sessions on your machines: a small agent on each box links it to the pitupitu cloud.
Claude-to-IM
claude-to-im
A bridge between Claude Code or Codex and messengers: talk to the agent from Telegram, Discord, Feishu, QQ or WeChat
README in short
The README lists features, explains the exchange scheme through the screen and files, requirements and installation for Windows, Linux and macOS. The move from the beta to the `_forever_` release client and the rename from wow-claude are described separately.