Projects that use an agent SDK (Anthropic, OpenAI, etc.) to build custom agents — with tools, hooks, subagents, MCP, and structured execution. The libraries themselves and the production agents that ship on top of them.
Self-improving AI agent platform — autonomously creates skills, learns from experience, runs anywhere
AI agent SDK for autonomous software development — self-hosted alternative to Devin and Jules
AI pair programming agent for your terminal — integrates with Claude, maps codebase, automates git commits
Coding agent CLI for unified workflows across OpenAI, Gemini, Ollama, and Claude Code models
Stealth Chromium browser for AI agents — bypasses bot detection with source-level fingerprint patches
Agent SDK application examples — quickstarts for autonomous coding, browser automation, and data analysis
Self-evolving agent framework for system control — grows skills to automate desktop, browser, and mobile tasks
Browser-based AI agent runtime — runs Claw Agents entirely in the browser with full Node.js environment
Autonomous Claude agent for GitHub — automates end-to-end development, code reviews, and PR management directly in your repos
Agent workspace manager for Claude Code — runs agents in isolated git worktrees and Docker containers
The Claude Agent SDK (@anthropic-ai/claude-agent-sdk) lets you embed Claude Code as a library in your own Python or TypeScript application. You get tools, hooks, subagents, and MCP support without running the full CLI.
Use an agent SDK when you’re building one agent that runs a task end-to-end. Use an orchestration framework (LangGraph, CrewAI) when you need multiple agents coordinating, with state machines or shared memory.
