How does this skill handle Claude's tool use?

It implements the mandatory tool-result loop, detecting stop_reason='tool_use' and providing patterns to execute tools and send results back to the model.

What is the 'extended thinking' feature?

Extended thinking allows Claude to use a specific token budget for internal reasoning before providing a final response, ideal for complex problem-solving.

Why is max_tokens always required in this skill?

The Anthropic API does not provide a default value for max_tokens, so it must be explicitly defined in every request to prevent API errors and manage costs.

Can I use this for multi-modal tasks?

Yes, it includes standard patterns for handling base64 images, URL images, and PDF/document processing via the Messages API content blocks.

Does it support streaming responses?

Yes, it provides both the .stream() helper for event-based text accumulation and raw Server-Sent Events (SSE) iteration for low-level control.

Anthropic SDK Implementation

Name: Anthropic SDK Implementation
Author: agents-inc

byagents-inc

•

Ciencia de Datos y ML

Implements production-grade patterns for the official Anthropic TypeScript SDK including Messages API, streaming, and tool use.

This skill provides standardized patterns and best practices for integrating the official Anthropic TypeScript/Node.js SDK into Claude Code projects. It covers everything from basic client configuration and the Messages API to advanced features like extended thinking, prompt caching, vision capabilities, and the Batch API. By enforcing strict adherence to mandatory token limits, robust error handling via Anthropic's specific error classes, and correct content block iteration, it ensures resilient and efficient interactions with Claude models while avoiding common implementation pitfalls like hardcoded keys or silent truncation.

Características Principales

01Advanced event-based streaming and raw SSE iteration patterns

02Support for Vision, PDF processing, and Extended Thinking reasoning blocks

03Type-safe Tool Use (function calling) and conversation loop management

045 GitHub stars

05Standardized Messages API setup for single and multi-turn conversations

06Production-ready error handling, automatic retries, and token counting

Casos de Uso

01Implementing structured data extraction using Zod schema validation

02Optimizing high-volume workflows with prompt caching and the Batch API

03Building LLM-powered applications using Claude Sonnet, Opus, or Haiku models

Características Principales

01Advanced event-based streaming and raw SSE iteration patterns

02Support for Vision, PDF processing, and Extended Thinking reasoning blocks

03Type-safe Tool Use (function calling) and conversation loop management

045 GitHub stars

05Standardized Messages API setup for single and multi-turn conversations

06Production-ready error handling, automatic retries, and token counting

Casos de Uso

01Implementing structured data extraction using Zod schema validation

02Optimizing high-volume workflows with prompt caching and the Batch API

03Building LLM-powered applications using Claude Sonnet, Opus, or Haiku models