Claude Code Proxy Configuration

This repository contains a shell script that provides intelligent proxy functionality for Claude Code CLI, allowing it to work with multiple AI model providers through a unified interface.

Overview

The .claude.sh script intercepts Claude Code CLI commands and automatically routes them to the appropriate model provider based on the --model parameter. It supports multiple Chinese AI providers while maintaining compatibility with the standard Claude Code interface.

Supported Model Providers

The script supports the following model providers and their corresponding models:

Provider	Environment Variable	Base URL	Supported Models
Zhipu (智谱)	ZHIPU_API_KEY	https://open.bigmodel.cn/api/anthropic	glm-4.5
Alibaba (阿里)	DASHSCOPE_API_KEY	https://dashscope.aliyuncs.com/api/v2/apps/claude-code-proxy	qwen3-coder-plus
DeepSeek	DEEPSEEK_API_KEY	https://api.deepseek.com/anthropic	deepseek-chat
Moonshot (月之暗面)	MOONSHOT_API_KEY	https://api.moonshot.cn/anthropic	kimi-k2-0905-preview
LongCat	LONGCAT_API_KEY	https://api.longcat.chat/anthropic	LongCat-Flash-Chat

Installation

1. Install Claude Code CLI:

   npm install -g @anthropic-ai/claude-code

2. Source the proxy script in your shell configuration:

   # Add to ~/.zshrc or ~/.bashrc
   source /path/to/.claude.sh

3. Set up your API keys for the desired providers:

   export ZHIPU_API_KEY=your_zhipu_key
   export DASHSCOPE_API_KEY=your_alibaba_key
   export DEEPSEEK_API_KEY=your_deepseek_key
   export MOONSHOT_API_KEY=your_moonshot_key
   export LONGCAT_API_KEY=your_longcat_key

Usage

Use Claude Code as normal, but specify the desired model with the --model parameter:

# Use Zhipu's GLM-4.5 model
claude --model glm-4.5

# Use Alibaba's Qwen3 Coder Plus
claude --model qwen3-coder-plus

# Use DeepSeek Chat
claude --model deepseek-chat

# Use Moonshot's Kimi K2
claude --model kimi-k2-0905-preview

# Use LongCat's model
claude --model LongCat-Flash-Chat

# Use default Claude model (no proxy)
claude

Features

• Automatic Provider Detection: The script automatically detects which provider to use based on the model name
• Environment Variable Validation: Checks that required API keys are set before routing requests
• Transparent Proxy: Maintains full compatibility with Claude Code CLI syntax and options
• Error Handling: Provides clear error messages when API keys are missing or invalid
• Flexible Model Parameter: Supports both --model model_name and --model=model_name formats

Configuration

The script uses a associative array (PROVIDERS) to configure provider mappings. Each entry contains:

• Environment variable name for the API key
• Base URL for the provider's Anthropic-compatible API
• Comma-separated list of supported models

Error Messages

The script provides helpful error messages:

• CLI Not Found: Instructions for installing Claude Code if not detected
• Missing API Key: Specific environment variable name needed for each provider
• Provider Routing: Confirmation when successfully routing to a specific provider

Timeout Configuration

All proxied requests use a 10-minute timeout (API_TIMEOUT_MS=600000) to accommodate longer processing times for complex tasks.

MCP Integration

This setup includes support for Model Context Protocol (MCP) servers to enhance AI assistant capabilities with additional tools and resources.

Available MCP Servers

• cf-docs: Cloudflare documentation search and resources
• context7: Up-to-date library documentation and code examples
• chrome-devtools: Browser developer tools integration

Configuration

For detailed MCP setup instructions and commands for different AI assistants (Claude, Gemini, Codex), see MCP.md.

MCP servers provide additional tools like:

• Access to current documentation for various libraries
• Cloudflare product documentation and resources
• Browser debugging and inspection capabilities
• Real-time code examples and API references