claude-code-router
Use Claude Code as the foundation for coding infrastructure, allowing you to decide how to interact with the model while enjoying updates from Anthropic.
open-sourceagent-frameworks
30.8k
Stars
+2010
Stars/month
0
Releases (6m)
Star Growth
+428 (1.4%)
Overview
Claude Code Router 是一个强大的路由工具，允许开发者以 Claude Code 为基础架构，将请求智能路由到不同的AI模型和提供商。该工具支持多达6个主要AI提供商（OpenRouter、DeepSeek、Ollama、Gemini、Volcengine、SiliconFlow），可根据任务类型（后台任务、深度思考、长上下文处理）动态选择最适合的模型。通过 `/model` 命令，用户可在 Claude Code 内实时切换模型，无需重启会话。该工具还提供了请求/响应转换器，支持自定义插件系统，并能与 GitHub Actions 集成，为CI/CD流程提供AI能力。拥有超过3万GitHub星标，是一个成熟且活跃的开源项目。
Deep Analysis
Key Differentiator
vs using Claude Code directly: Route different types of requests (background, thinking, long context) to optimal models from any provider, saving costs while maintaining Claude Code's UX - with dynamic switching and plugin system
⚡ Capabilities
- • Route Claude Code requests to different LLM models
- • Multi-provider support (OpenRouter, DeepSeek, Ollama, Gemini, etc.)
- • Request/response transformation with plugin system
- • Dynamic model switching via /model command
- • CLI model management
- • GitHub Actions integration
- • Preset configuration management
- • Web UI for configuration
🔗 Integrations
Claude CodeOpenRouterDeepSeekOllamaGoogle GeminiVolcengineSiliconFlowModelScopeDashScopeGitHub Actions
✓ Best For
- ✓ Claude Code users wanting to use cheaper/alternative models for different task types
- ✓ Teams optimizing LLM costs by routing background tasks to cheaper models
- ✓ Developers needing multi-provider flexibility within Claude Code
✗ Not Ideal For
- ✗ Users not using Claude Code
- ✗ Teams that only use Anthropic models
Languages
TypeScript/JavaScript
Deployment
npm install -gLocal proxy server
Pricing Detail
Free: Fully free and open-source
Paid: N/A - bring your own model API keys
⚠ Known Limitations
- ⚠ Only works with Claude Code CLI
- ⚠ Requires separate LLM API keys for each provider
- ⚠ Proxy approach may introduce latency
- ⚠ Configuration requires understanding of model provider APIs
Pros
- + 支持6个主要AI提供商的无缝切换，可根据任务需求选择最合适的模型
- + 提供动态模型切换和CLI管理功能，操作简便且支持实时调整
- + 可扩展的插件系统和请求转换器，允许深度定制和与现有工作流集成
Cons
- - 需要依赖 Claude Code 作为基础框架，增加了环境配置复杂性
- - 需要手动配置多个提供商的API密钥和参数设置
- - 作为中间层可能引入额外的延迟和潜在的故障点
Use Cases
- • AI开发团队需要根据不同任务类型（编码、分析、创作）使用不同模型的场景
- • 希望在GitHub Actions中集成多个AI提供商能力的CI/CD自动化流程
- • 需要灵活切换AI模型以优化成本和性能的企业级AI应用开发
Getting Started
1. 安装依赖：先安装Claude Code (`npm install -g @anthropic-ai/claude-code`)，再安装路由器 (`npm install -g @musistudio/claude-code-router`)；2. 配置设置：创建 `~/.claude-code-router/config.json` 文件，添加各AI提供商的API密钥和路由规则；3. 开始使用：在Claude Code中使用 `/model` 命令切换模型，或通过 `ccr model` CLI命令管理模型配置。