Agent using the Anthropic SDK directly. Claude models with tool use and streaming — no framework overhead, just the official SDK.
What's included
- Direct Anthropic SDK — no abstraction layer, use the official client
- Claude models — Claude 3.5 Sonnet, Claude 3 Opus, and more
- Tool use — define tools and let Claude invoke them
- Streaming — real-time token streaming via SSE
- Full control — direct access to system prompts and parameters
When to use this
Pick this example when you want to:
- Use the Anthropic SDK directly without any framework
- Build agents with Claude models
- Have full control over prompts and tool definitions
Project structure
├── agents/
│ └── anthropic_agent.py # Anthropic SDK agent
├── main.py # Entry point — serve(agents)
├── pyproject.toml
└── reminix.config.toml
Quick start
Deploy the example and invoke the agent via the Reminix API. The Anthropic SDK handles chat completions, tool use, and streaming — you get a production REST endpoint with direct Claude access.