What problem does it solve?
This Skill provides a standardized way to send conversational turns to Claude and receive structured, block-based responses so applications can power chat, completion, analysis, and multimodal workflows without bespoke request plumbing.
Core Features & Use Cases
- Unified Messages API: Send multi-turn conversations including text, images, documents, tool results, and system prompts to get the next assistant response.
- Streaming & Synchronous Modes: Support for both synchronous create calls and token-by-token SSE streaming for real-time UIs.
- Structured Outputs & Tooling: Built-in support for tool definitions, structured JSON output schemas, extended thinking, and detailed usage/stop_reason metadata for robust integration.
- Use Case: Build a chat interface that streams Claude's tokens to the user, handles tool calls mid-conversation, and validates the final response against a JSON schema for downstream processing.
Quick Start
Send a chat message to Claude asking "Hello, Claude" using model claude-sonnet-4-6 with max_tokens 1024 and read the assistant text from the returned content block.