A comprehensive development framework for building LLM-powered applications using the Anthropic Claude API and the specialized Agent SDK.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install claude-api-cn
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install claude-api-cn using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Claude API & Agent SDK Integration?
This Openclaw Skills module provides a robust architecture for developing applications powered by Claude. It enables developers to integrate advanced features like adaptive thinking, server-side code execution, and high-fidelity structured outputs across a wide variety of programming environments. By bridging the gap between simple API calls and complex autonomous agents, it allows for scalable AI integration tailored to specific project needs.
The skill emphasizes the use of Claude Opus 4.6 as the flagship model, supporting context windows up to 1M tokens in beta and introducing adaptive thinking. This ensures that complicated reasoning tasks are handled with dynamic token allocation, while the Agent SDK offers out-of-the-box support for file systems, web browsing, and terminal access, making it a premier choice for agentic coding assistants and research tools.
Claude API & Agent SDK Integration Use Cases
- Building intelligent classification and data extraction pipelines.
- Developing agentic coding assistants with direct file and terminal access.
- Implementing multi-step workflows with automated tool-use loops.
- Generating formatted reports (PDF, DOCX) and data visualizations using server-side code execution.
- Creating real-time chat interfaces with high-performance streaming and adaptive reasoning.
How Claude API & Agent SDK Integration Works
- The system performs automatic language detection by scanning project manifest files like package.json or requirements.txt.
- It selects the optimal integration tier: a single API call for simple tasks, a tool-use workflow for orchestrated logic, or the Agent SDK for autonomous exploration.
- Requests are routed through the Messages API with specialized configurations for adaptive thinking and output format.
- For long-running sessions, server-side compaction is triggered to summarize context and maintain performance within token limits.
- The SDK tool runner manages the execution loop, handling tool calls and feeding results back to the model until the task is complete.
Claude API & Agent SDK Integration Setup
First, ensure you have your Anthropic API key configured. Install the relevant SDK for your environment:
# For Python projects
pip install anthropic
# For TypeScript/JavaScript projects
npm install @anthropic-ai/sdk
Initialize the client and enable adaptive thinking for complex reasoning:
from anthropic import Anthropic
client = Anthropic()
message = client.messages.create(
model="claude-opus-4-6",
max_tokens=16000,
thinking={"type": "adaptive"},
messages=[{"role": "user", "content": "Analyze this codebase."}]
)
Claude API & Agent SDK Integration Data Schema & Taxonomy
The skill organizes data around the Messages API architecture and language-specific SDK implementations:
| Component | Description |
|---|---|
| Model ID | Uses static identifiers like claude-opus-4-6 or claude-sonnet-4-6. |
| Output Config | Defines effort levels (low, medium, high, max) and structured format constraints. |
| Tool Schema | Utilizes Zod (TS) or Decorators (Python) to define function metadata and parameter validation. |
| Compaction State | Preserves response.content blocks to maintain summarized context across turns. |
| File API | Manages persistent file uploads for cross-request reference without re-uploading. |
Claude API & Agent SDK Integration Advanced Features
- Adaptive Thinking: Dynamically scales reasoning depth without requiring deprecated manual token budgets.
- Server-Side Compaction: Automatically manages context windows by summarizing history at the 150K token threshold.
- Multi-Language Tool Runners: Beta support for automated loop handling in Python, TypeScript, Java, Go, and Ruby.
- Native MCP Support: Seamless integration with the Model Context Protocol via the Agent SDK.
- Advanced File Generation: Native support for generating DOCX, PDF, and Matplotlib charts within the execution sandbox.
SKILL.md
Loading
Related Openclaw Skills
Find Skills
A specialized discovery tool for listing and searching through all installed AI agent capabilities within a workspace.
Wudao Intel
A comprehensive A-share market intelligence skill providing real-time data on dragon-tiger boards, institutional research, and pre-market auction statistics.
Chat Logger
A highly reliable chat recording and summarization tool designed for Feishu and DingTalk that utilizes strict trigger rules to ensure accurate data capture.
Feedback Page Generator
A specialized tool for designing, optimizing, and auditing user feedback and product roadmap pages to drive community engagement.
App Ads Skill
A comprehensive guide for executing high-performing mobile app install campaigns and user acquisition strategies across major ad platforms.
DOCX Document Master
A professional-grade toolkit for programmatically creating, editing, and analyzing Microsoft Word documents with high-fidelity formatting.
Featured*
