llm-api-builder

Build applications with the Claude API, Anthropic SDKs, and Agent SDK across multiple languages.

5|Updated Jun 13, 2026
One-click install
npx skills add https://github.com/PatriotAi/ai-lab --skill llm-api-builder-patriotai
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: llm-api-builder
Source: https://github.com/PatriotAi/ai-lab/tree/main/melania-skills-ecosystem/skills/llm-api-builder
Command: npx skills add https://github.com/PatriotAi/ai-lab --skill llm-api-builder-patriotai

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? Developers integrating the Claude API or Anthropic SDKs face scattered documentation, model version drift, and subtle pitfalls like broken prompt caching, deprecated thinking parameters, and request timeouts on long outputs. This Skill provides a single entry point with surface selection guidance, language detection, and current best practices for building LLM-powered applications. ## Core Features & Use Cases - Surface & Language Detection: Automatically picks the right integration surface (Claude API, Agent SDK, Batch API) and loads language-specific guidance for Python, TypeScript, Java, Go, Ruby, C#, PHP, or cURL. - Advanced API Patterns: Covers streaming, tool use, structured outputs, prompt caching, citations, vision, PDF support, Files API, Batch API, compaction, and the memory tool with concrete code examples. - Pitfall Prevention: Flags deprecated parameters (e.g., budget_tokens on current flagship models), cache-breaking patterns, and timeout risks before they cause production issues. - Use Case: A developer asks to add streaming and tool use to a Node.js backend; the Skill detects TypeScript, selects the Claude API surface, and provides current SDK patterns with adaptive thinking and structured outputs. ## Quick Start Ask the assistant to build an application using the Claude API with tool use and streaming in your project's programming language.

Frequently Asked Questions about llm-api-builder

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build an app with the Claude API and tool use?▼

Define tools via decorators, Zod schemas, or raw JSON, then let the SDK tool runner execute your functions and loop until Claude finishes. The Skill detects your project language and loads the matching guide for Python, TypeScript, Java, Go, or Ruby.

Claude API vs Agent SDK: which should I use?▼

Use the Claude API for single calls, classification, extraction, or custom workflows where you control the loop. Use the Agent SDK when the agent needs built-in file, web, or terminal access with permissions and guardrails included.

Does the Anthropic SDK support streaming for long responses?▼

Yes, streaming is the default recommendation for requests with long inputs, long outputs, or high max_tokens to avoid request timeouts. Use the SDK helper .get_final_message() or .finalMessage() to collect the complete response without handling individual stream events.

Why is budget_tokens not working with current Claude models?▼

The budget_tokens parameter is deprecated on current flagship models, which use adaptive thinking via thinking: {type: "adaptive"} instead. The parameter remains valid only for older model generations; verify current behavior in the official docs.

How do I reduce token costs with the Claude API?▼

Use prompt caching for repeated context (up to 90% savings on cached tokens), Batch API for asynchronous bulk requests (50% discount), and Tool Search Tool or Programmatic Tool Calling for large tool libraries. Keep system prompts and tool definitions stable to preserve cache hits.

Can I use this Skill with the OpenAI SDK or other AI providers?▼

No, the Skill explicitly does not trigger for openai or other non-Anthropic SDK imports. Its scope is limited to the Claude API, Anthropic SDKs, and the Claude Agent SDK.