Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn to build autonomous AI agents using Anthropic's Claude Agent SDK through this comprehensive workshop that progresses from theoretical foundations to hands-on implementation. Explore the evolution from simple LLM features to complex workflows and finally to autonomous agents that build their own context and decide their trajectories independently. Master the core "Harness" architecture that encompasses tools, prompts, file systems, skills, sub-agents, and memory components essential for robust agent development. Follow along with live coding demonstrations as you implement the fundamental Agent Loop pattern of Context → Thought → Action → Observation, integrating the powerful Bash tool for general computer use that leverages existing software like ffmpeg, grep, and git. Discover advanced Context Engineering techniques using the file system to maintain state across long tasks, where files serve as both memory and verification mechanisms for agent actions. Understand safety considerations through ReadOnly vs ReadWrite file access permissions and learn to implement monitoring systems that provide real-time visibility into the agent's thought processes. Address common challenges including handling "stuck" states through feedback loops and error correction mechanisms, while exploring multi-turn complex tasks through practical research agent demonstrations. Gain insights into refactoring patterns using "Hooks" for deterministic overrides that can intercept and correct agent behavior without model retraining. Participate in detailed Q&A sessions covering reproducibility strategies, approaches for managing massive codebases exceeding 50 million lines of code, and techniques for overcoming semantic search limitations in large-scale applications.
