codelean

codelean is a faster, cheaper, safer Codex-like coding agent with a multi-provider model boundary. It keeps the provider layer small so you can switch cloud model backends without changing caller code.

Current Scope

• Use a Codex fork as the future execution base.
• Optimize for lower cost, lower latency, and safer execution.
• Keep Codex-like file/shell/approval/diff/test capabilities as the product direction.
• Provide a unified ModelProvider boundary for model calls, streaming, tool-call events, usage, and errors.
• Support anthropic directly and OpenAI-compatible providers including openai, qwen, and mimo.
• Keep gemini planned but not implemented yet.

Install

npm install
npm run build

Usage

MiMo

MiMo is OpenAI-compatible. The default MiMo base URL is https://api.xiaomimimo.com/v1, and the default model is mimo-v2.5-pro.

export MIMO_API_KEY="sk-..."
node dist/cli/index.js --provider mimo --prompt "Explain this repo"

You can also set defaults:

export CODELEAN_PROVIDER="mimo"
export CODELEAN_MODEL="mimo-v2.5-pro"
export MIMO_API_KEY="sk-..."
node dist/cli/index.js --prompt "Write a TypeScript function"

For Token Plan usage, override the base URL:

export MIMO_BASE_URL="https://token-plan-cn.xiaomimimo.com/v1"

Anthropic

export ANTHROPIC_API_KEY="sk-ant-..."
node dist/cli/index.js --provider anthropic --model claude-sonnet-4-5 --prompt "Explain this repo"

OpenAI-compatible providers

export OPENAI_API_KEY="sk-..."
node dist/cli/index.js --provider openai --model gpt-4o --prompt "Explain this repo"

export QWEN_API_KEY="sk-..."
node dist/cli/index.js --provider qwen --model qwen-plus --prompt "Explain this repo"

Environment

| Variable | Description | |----------|-------------| | CODELEAN_PROVIDER | Default provider: anthropic, openai, qwen, or mimo | | CODELEAN_MODEL | Default model for the selected provider | | CODELEAN_API_KEY | Generic API key override for the selected provider | | CODELEAN_BASE_URL | Generic base URL override for OpenAI-compatible providers | | ANTHROPIC_API_KEY | Anthropic API key | | OPENAI_API_KEY | OpenAI API key | | QWEN_API_KEY | Qwen API key | | MIMO_API_KEY | MiMo API key | | MIMO_BASE_URL | MiMo base URL, default https://api.xiaomimimo.com/v1 | | CODELEAN_MAX_OUTPUT_TOKENS | Max output tokens, default 4096 |

Code Structure

src/
  cli/index.ts                          # Minimal CLI
  client.ts                             # CodeleanClient
  config.ts                             # Multi-provider config
  providers/anthropic-provider.ts       # Anthropic Messages API adapter
  providers/openai-compatible-provider.ts # OpenAI-compatible adapter base
  providers/mimo-provider.ts            # MiMo adapter
  providers/provider-factory.ts         # Provider selection
  types.ts                              # Shared request/response/event types

Next Steps

1. Add provider-specific tests with mocked HTTP responses.
2. Add a real GeminiProvider implementation.
3. Add fallback routing between primary and secondary providers.
4. Map Codex internal tool calls into the unified provider boundary.
5. Integrate this provider layer into a Codex fork while preserving Codex approval and sandbox behavior.