Feb 24
Loading video player...
Agent harnesses are the future of agentic coding and how we can build our coding agent into a full blown AI engineer. Just like a human engineer, we need it to manage issues, create branches, open PRs, notify you when things are done, pick up the next task automatically, etc. That's what I've been experimenting with using subagents, the Anthropic harness for long-running agents, and Arcade's MCP gateway. I've taken my Claude Code to the next level - it tracks progress in Linear, works with repos in GitHub, and keeps me updated in Slack. All through subagents that handle the context management for each service. ~~~~~~~~~~~~~~~~~~~~~~~~~~ - Check out Arcade, the platform for agent authorization that made MCP enterprise-ready: https://arcade.dev.plug.dev/wNi0ZaK ~~~~~~~~~~~~~~~~~~~~~~~~~~ - The Dynamous Agentic Coding Course is now FULLY released - learn how to build reliable and repeatable systems for AI coding: https://dynamous.ai/agentic-coding-course - AI Engineer Repository (Your Claude Engineer): https://github.com/coleam00/your-claude-engineer - Excalidraw Diagram for the Harness: https://github.com/coleam00/your-claude-engineer/blob/main/AgentHarnessDiagram.png - Anthropic Long-Running Agent Harness: https://www.anthropic.com/engineering/effective-harnesses-for-long-running-agents ~~~~~~~~~~~~~~~~~~~~~~~~~~ 0:00 Introduction to Agent Harnesses 1:35 Building an AI Engineer (How it Works) 5:35 Setting Up Our AI Engineer Harness 8:06 Running the Harness Live 10:11 How the Agent Harness Works (Diagram) 14:15 Customizing Prompts and Subagents 17:33 Final Results and App Review 18:34 Future of AI Coding Workflows ~~~~~~~~~~~~~~~~~~~~~~~~~~ Join me as I push the limits of what is possible with AI. I'll be uploading videos weekly - at least every Wednesday at 7:00 PM CDT!