CommandCode Proxy Server

Overview

CommandCode Proxy Server is an open-source proxy server, acting as a bridge between AI coding tools (Claude Code, Codex, Kilo Code, Open Code...) and CommandCode.AI — via the OpenAI API standard.

To put it simply: this repo allows you to use any AI coding tool that supports the OpenAI API to call the AI models of CommandCode.AI without needing to change the code of that tool.

The Problem

Currently, CommandCode.AI offers Plan Go for only $1/month but comes with a usage quota equivalent to $10. However, this quota can only be used within the CommandCode coding agent — it does not allow direct API calls from the outside.

The Solution

This repo acts as an intermediary proxy:

• Receives standard OpenAI requests from the client side

• Converts to CommandCode format

• Sends to upstream https://api.commandcode.ai/alpha/generate

• Receives response from CommandCode and converts it back to standard OpenAI format

Main Features

OpenAI API compatibility

• /v1/chat/completions — standard OpenAI chat completions endpoint

• /chat/completions — alias without the v1 prefix

• /v1/responses — compatible with OpenAI Responses API

• /v1/models — returns a list of models

Streaming

• Fully supports streaming and non-streaming

• Converts SSE (Server-Sent Events) from CommandCode to OpenAI format

Tool / Function Calling

• Converts tool definitions from OpenAI to CommandCode schema

• Fully processes tool events from upstream:

  • tool-use

  • tool-delta

  • tool-input-start

  • tool-input-delta

  • tool-call

• Coding tools can execute tools and return the results

Model aliases

Supports multiple models with concise aliases:

AliasModelkimi-k2.6moonshotai/Kimi-K2.6glm-5.1zai-org/GLM-5.1minimax-m2.7MiniMaxAI/MiniMax-M2.7deepseek-v4-prodeepseek/deepseek-v4-proqwen-3.6-plusQwen/Qwen3.6-Plusstep-3.5-flashstepfun/Step-3.5-Flashgemini-3.1-flash-litegoogle/gemini-3.1-flash-lite

How to Use

# Run server
go run main.go

# With API key
go run main.go -api-key YOUR_COMMANDCODE_API_KEY

# On a custom port
go run main.go -port 8080

# Build for Linux
GOOS=linux GOARCH=amd64 go build -o bin/command-code-proxy

Then direct your AI coding tool to http://127.0.0.1:55990 instead of https://api.openai.com.

Example Request

curl http://127.0.0.1:55990/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_COMMANDCODE_API_KEY" \
  -d