# mimo2codex
<p align="center">
<a href="./README.md"><strong>English</strong></a> ·
<a href="./README.zh.md">简体中文</a> ·
<a href="https://mimodoc.chengj.online">Docs site</a>
</p>
<p align="center">
<a href="https://github.com/7as0nch/mimo2codex/stargazers"><img alt="GitHub Stars" src="https://img.shields.io/github/stars/7as0nch/mimo2codex?style=flat-square&logo=github"></a>
<a href="https://www.npmjs.com/package/mimo2codex"><img alt="npm version" src="https://img.shields.io/npm/v/mimo2codex?style=flat-square&logo=npm"></a>
<a href="https://www.npmjs.com/package/mimo2codex"><img alt="downloads" src="https://img.shields.io/npm/dt/mimo2codex?style=flat-square&color=brightgreen"></a>
<img alt="license" src="https://img.shields.io/github/license/7as0nch/mimo2codex?style=flat-square">
<img alt="node" src="https://img.shields.io/badge/Node-18%2B-blue?style=flat-square&logo=node.js&logoColor=white">
<img alt="wire_api" src="https://img.shields.io/badge/wire__api-responses-black?style=flat-square">
<a href="https://paypal.me/7as0nch"><img alt="Sponsor on PayPal" src="https://img.shields.io/badge/Sponsor-PayPal-00457C?style=flat-square&logo=paypal&logoColor=white"></a>
</p>
Local proxy that lets the **latest OpenAI Codex CLI / desktop** talk to virtually any modern LLM. Built-in support for **Xiaomi MiMo V2.5** and **DeepSeek V4 Pro**, plus a **generic provider mechanism** for any **OpenAI Chat Completions-compatible** (Qwen / GLM / Kimi / vLLM / Ollama / LM Studio …) or **native Responses API** upstream — no code changes, no re-publish. It translates Codex's Responses API ↔ upstream Chat Completions on the fly, routes per-request by the `model` field, and runs on `127.0.0.1`.
**Why:** MiMo's official Codex integration only supports `wire_api = "chat"`, which newer Codex versions hard-error on. mimo2codex sits in between, so you keep Codex on latest and it thinks it's talking to a native Responses backend. Conceptually a thin protocol shim — sibling to [openrouter](https://openrouter.ai) / [claude-code-router](https://github.com/musistudio/claude-code-router).
## Three ways to run
1. **CLI** — `npm install -g mimo2codex`. The classic path; full control from the terminal.
2. **Docker** — for intranet / team setups: user login, BYOK, OAuth, downloadable Codex config bundles; the upstream key never leaks. → [Auth & deployment](./doc/auth-deployment.md)
3. **Desktop app** (Windows / macOS, best for non-technical users) — installer, runs in the background, starts on boot; one click from the tray opens the admin UI. → [Download](https://mimodoc.chengj.online/download)
## What it does
- ✅ **Codex CLI (`wire_api = "responses"`) + desktop app** — stay on the latest Codex.
- ✅ **Multi-provider in one process** — MiMo + DeepSeek + generic providers, per-request routing by the `model` field.
- ✅ **Any OpenAI-compatible / native-Responses upstream** — Qwen / GLM / Kimi / Ollama / OpenAI, declared in `providers.json`.
- ✅ **Tool calling, web search, vision, reasoning** — function & parallel tools, MCP namespace; MiMo native `web_search`; correct multi-turn `reasoning_content` round-trip.
- ✅ **One-click Codex model switching** from the admin webui (replaces cc-switch).
- ✅ **Admin console** at `http://127.0.0.1:8788/admin/` — model catalog, chat logs, token stats, provider config; sqlite persistence.
## Quick start
```bash
npm install -g mimo2codex # 1. install (Node ≥ 18)
mimo2codex init # 2. add your API key → edit ~/.mimo2codex/.env
mimo2codex # 3. start the proxy on 127.0.0.1:8788
```
Then point Codex at it — one-click on the admin UI's **Codex Enable** page, or copy the printed `config.toml` / `auth.json` into `~/.codex/`. Walkthrough: **[Codex Enable](./doc/codex-enable.md)**.
> Keys & per-OS setup → [Env setup](./doc/env-setup.md) · Run Codex CLI in isolation on Windows → [Isolated Codex CLI](./doc/codex-cli-isolated-windows.md)
## Documentation
Full docs (searchable, bilingual) live at **<https://mimodoc.chengj.online>**.
**Getting started**
- [Env setup](./doc/env-setup.md) — set up all keys once; per-OS `.env` loader (macOS / Linux / Windows)
- [Codex Enable](./doc/codex-enable.md) — one-click model switching in the webui (replaces cc-switch)
- [Isolated Windows Codex CLI](./doc/codex-cli-isolated-windows.md) — let Codex CLI use MiMo while Codex Desktop stays untouched
**Deployment**
- [Docker](./doc/docker.md) — `docker compose up -d`, data persistence, multi-arch images
- [Auth & multi-user](./doc/auth-deployment.md) — login, BYOK, OAuth, downloadable Codex config bundles
**Providers**
- [Generic providers](./doc/generic-providers.md) — any OpenAI-compatible / native-Responses upstream via `providers.json`
- [MiniMax](./doc/minimax.md) · [SenseNova](./doc/sensenova.md) · [Kimi](./doc/kimi.md) — per-vendor compatibility notes
- [Connector plugins](./doc/connector-plugins.md) — why Codex Desktop connectors (GitHub / Gmail / …) can't be proxied, and the fallback
**Reference**
- [mimoskill](./doc/mimoskill.md) — image gen / OCR fallback / `/hatch` pet generation
- [Proxy & network FAQ](./doc/proxy-faq.md)
- [Community feedback](./doc/community-feedback.md)
- [Tag log (changelog)](./doc/tag-log.md)
## License
MIT — see [LICENSE](./LICENSE).