anthropic-proxy
🥈SilverProxy server that converts Anthropic API requests to OpenAI format for use with OpenRouter. Enables operations teams to use Claude Code with OpenRouter instead of the Anthropic API. Connects Anthropic's API format to OpenAI-compatible endpoints.
415330Updated 2mo ago
Intermediate30min to implementautomation
Saves ~60 min per use
Quick InstallView Source
git clone https://github.com/maxnowack/anthropic-proxy.gitWorks with:
Claude
Overview
About This Skill
anthropic-proxy is a proxy server that converts Anthropic API requests into OpenAI-compatible format and routes them to OpenRouter or custom endpoints. It enables operations teams and developers to use Claude Code with OpenRouter instead of the Anthropic API directly. The proxy listens on a configurable port (default 3000) and transforms requests transparently, allowing you to maintain Anthropic's API format while connecting to OpenAI-compatible services. Configure it with environment variables for your OpenRouter API key, custom base URLs, model selection, and port settings. This is useful for teams needing gateway flexibility, custom routing, or integration with OpenAI-compatible infrastructure.
How to Use
[{"step":"Install the proxy using the provided Docker image or Helm chart. Run `docker pull ghcr.io/your-org/anthropic-proxy:latest` and configure environment variables for OPENROUTER_API_KEY and ANTHROPIC_API_KEY.","tip":"Use Docker Compose for local testing with volumes mapped to your config directory. Set `DEBUG=true` for verbose logging during setup."},{"step":"Configure your application to point to the proxy endpoint instead of Anthropic's API. Replace `api.anthropic.com` with `localhost:8080` (or your proxy URL) in your Claude Code client initialization.","tip":"Test with `curl -X POST http://localhost:8080/v1/messages -H 'Content-Type: application/json' -d '{\"model\":\"claude-3-opus\",\"messages\":[{\"role\":\"user\",\"content\":\"Hello\"}]}'` to verify the proxy is responding."},{"step":"Validate the proxy conversion by comparing Anthropic's raw response with the OpenRouter output. Use the example output format to ensure metadata like token counts are preserved.","tip":"Automate validation with a GitHub Actions workflow that runs `anthropic-proxy-test` after each deployment to catch format regressions."},{"step":"Deploy to production with load balancing. Configure the proxy with horizontal scaling (e.g., Kubernetes HPA) to handle traffic spikes during peak usage.","tip":"Monitor the `/health` endpoint and set up alerts for proxy response times exceeding 500ms."}]
Use Cases
Integrate Claude Code into existing applications that rely on Anthropic API without significant rewrites.
Automate the transformation of API requests to streamline workflows between Anthropic and OpenAI services.
Use the proxy server to test and develop applications that require OpenAI-compatible endpoints while using Anthropic's API format.
Enable debugging and logging of API requests to troubleshoot issues in real-time during development.
Tags
Setup & Installation
Quick Install
No install command available. Check the GitHub repository for manual installation instructions.
Alternative Install (Git Clone)
git clone https://github.com/maxnowack/anthropic-proxyRequirements
- Claude Code or compatible AI agent
- Works with: Claude
Quick Start Guide
1
Install the Skill
Copy the install command above and run it in your terminal.
2
Open Your AI Agent
Launch Claude Code, Cursor, or your preferred AI coding agent.
3
Try It Out
Use the prompt template or examples below to test the skill.
4
Customize
Adapt the skill to your specific use case and workflow.
Usage Examples
Prompt Template
Set up an Anthropic-to-OpenAI proxy for [OPENROUTER_ENDPOINT] to enable Claude Code usage. Configure the proxy to handle [API_VERSION] requests and ensure compatibility with [ANTHROPIC_MODEL_NAME]. Test the proxy with a sample prompt: '[SAMPLE_PROMPT]' and verify the response format matches OpenAI's structure. Document the proxy configuration in [CONFIG_FILE_PATH].
Example Output
```json
{
"id": "cmpl-proxy-12345",
"object": "chat.completion",
"created": 1712345678,
"model": "claude-3-opus-openrouter",
"choices": [
{
"index": 0,
"message": {
"role": "assistant",
"content": "I've successfully configured the Anthropic-to-OpenAI proxy for OpenRouter. The proxy now accepts requests in Anthropic's format and converts them to OpenAI-compatible responses. The sample prompt 'Explain quantum computing in simple terms' was processed correctly, returning a structured response with the expected metadata. Configuration file saved to /etc/anthropic-proxy/config.json with endpoint settings for claude-3-opus-openrouter."
},
"finish_reason": "stop"
}
],
"usage": {
"prompt_tokens": 25,
"completion_tokens": 89,
"total_tokens": 114
}
}
```
**Proxy Configuration Summary:**
- **Source Model:** claude-3-opus-20240229
- **Target Endpoint:** https://openrouter.ai/api/v1
- **Conversion Rules Applied:**
- Anthropic's `messages` array → OpenAI's `messages` array
- `system` parameter → `messages[0].content` with role 'system'
- `max_tokens` → `max_completion_tokens`
- **Validation Results:**
- ✅ API request/response format conversion successful
- ✅ Token usage metrics preserved
- ✅ Error handling for unsupported Anthropic parameters implemented
**Next Steps:**
1. Deploy the proxy to your Kubernetes cluster using the provided Helm chart.
2. Update your CI/CD pipeline to route Claude Code requests through this proxy.
3. Monitor performance with Prometheus metrics exposed at `/metrics`.Apply to these tools
Browse all toolsOpenAI
Pioneering accessible, high-performance AI models
Claude
AI assistant built for thoughtful, nuanced conversation
Codility
Streamline tech recruiting with automated coding assessments and customizable tests.
IronCalc
IronCalc is a spreadsheet engine and ecosystem
Bonusly
Foster employee recognition with customizable rewards and real-time peer acknowledgments.
Microsoft Teams
Get more done every day with Microsoft Teams – powered by AI
Compatible MCP servers
Browse all MCP servers
Find the right skills for your stack
Take a free 3-minute scan and get personalized AI skill recommendations.
Take free scan