GenAIcode CLI
GenAIcode gives one API and CLI over Claude Code, Codex, Copilot, Cursor, Gemini CLI and opencode, with a verify-and-repair loop
- 65 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- Drivers launch coding agents that edit files and run commands, and `yolo` mode removes confirmations
- Agents get tasks, MCP servers and a permission policy from the call, so a settings mistake widens their access
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/genaicodeDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Drivers launch coding agents that edit files and run commands, and `yolo` mode removes confirmations
- Agents get tasks, MCP servers and a permission policy from the call, so a settings mistake widens their access
This is third-party code. Review the repository files before installing.
What it does
GenAIcode is an npm package that puts the coding agents you already have behind one interface: Claude Code, Codex, Copilot CLI, Cursor, Gemini CLI, opencode, Muse, Mistral Vibe and Antigravity. The agent keeps its own login, billing and sandbox, and GenAIcode picks no model and adds no agent loop of its own. On the command line `npx genaicode` opens a chat with the agent in the terminal, `npx genaicode ui` the same chat in the browser and `npx genaicode agents` lists the agents it finds. `npx genaicode run --verify "npm test" "fix the failing tests"` runs the first agent it finds, then your check, and a failing check goes back to the agent for another try. `-a codex` or `claude` picks the agent while the command stays the same. `npx genaicode embed` asks your agent to wire GenAIcode into your app. As a TypeScript library `genaicode/agents` gives every driver the same events and result, supports steering a running task, approvals and sandbox policy (`permissions` or `'yolo'`) and per-task MCP servers. For apps there are `genaicode/vite`, `genaicode/react`, `genaicode/app-agent` and `genaicode acp` for editors and game engines. Node.js 20 or newer is required.
Who it is for. Developers who use several coding agents and want one command to drive them, and app authors embedding an agent into their product.
Good fit when
- You need `npm test` to run after the agent works and the failure to go back to it for a retry automatically
- The same task has to run through Claude Code or Codex just by changing `-a`
- A Vite dev server needs a panel where an agent edits your app
- An editor or game engine has to attach an agent through the ACP protocol with `genaicode acp`
Not a fit when
- You need a built-in agent loop of its own: version 2 only wraps other agents and the old agent lives on the `1.x` branch
- None of the supported agents is on the machine: you install them and sign in yourself
- You need strict control over actions: `yolo` mode removes confirmations
Example request
Run genaicode run with the npm test check and have the agent fix the failing tests, retrying if the check failsLimitations
- GenAIcode installs no agents: you install them and sign in yourself, and `npx genaicode agents` shows what it found.
- Node.js 20 or newer is required; importing `genaicode` spawns nothing and `genaicode/agents` loads no provider SDK or UI code.
- Version 2 dropped the built-in agent: the old 1.x lives on a branch of the same name and runs with `npx genaicode@1`.
- Permission policy and `yolo` mode are set per call; in `yolo` mode agents act without confirmation.
How to disable. Remove the package with `npm uninstall genaicode`; run through `npx` it leaves nothing on the system except the npm cache.
FAQ
How do I get started with GenAIcode?
Run `npx genaicode` in a project terminal. `npx genaicode ui` opens the same chat in a browser and `npm install genaicode` adds it to code.
How do I make an agent retry until tests pass?
Use `npx genaicode run --verify "npm test" "fix the failing tests"`: after the agent works the check runs, and a failure goes back to the agent.
Which agents are supported?
Claude Code, Codex, Copilot CLI, Cursor, Gemini CLI, opencode, Muse, Mistral Vibe and Antigravity. The `-a` flag picks one, for example `-a codex`.
Does GenAIcode call models directly?
No. The agent keeps its own login, billing and sandbox, and GenAIcode picks no model and adds no loop of its own. `genaicode/providers` is a thin LLM layer.
How do I embed an agent in my app?
Run `npx genaicode embed` and the agent wires GenAIcode into your project. By hand use `genaicode/react` for a chat, `genaicode/vite` for a panel and `genaicode/app-agent`.
Discussion
No questions or reviews yet.
Related
CLI toolsA teaching terminal coding agent in Python, built as a "nanoGPT for agents": read it, take it apart and fork it
An open coding agent for the terminal, browser and desktop that works with models from OpenAI and Claude to Ollama
A plugin that gives an agent the IntelliNode skill and MCP server: ask any model, search, review code, tests, SQL, images and speech
A TypeScript multi-agent orchestrator with a locking shared blackboard, token budgets and an audit trail.
A Go CLI that configures the AI coding agents you already use: it gives them shared memory, a workflow and deterministic steps without replacing the agents them
OpenAI's coding agent that runs locally in your terminal, with IDE and desktop versions
README in short
The README gives a quick start with four commands, a table of uses from the CLI to an embedded agent, a TypeScript example with the Codex driver and a note on moving from 1.x.