Agent Harness Fundamentals
Claude Code can explore repositories, write and modify code, run commands, and take on increasingly complex development tasks. But as AI coding agents take on more responsibility, developers need to think beyond what the model can do.
How do you control the context it receives? Which capabilities should it use? When should work be delegated? And how do you verify the result before accepting it?
In this free 60-minute live webinar, Youssef Hosni and Luca Berton will show you how to build a practical agent harness around Claude Code, bringing together context, Agent Skills, specialist subagents, hooks, tests, guardrails, and human review into one structured workflow. Free 60 min Webinar Luca + Yous…
You’ll see the pieces come together live, from defining reusable capabilities to adding deterministic checks and specialist verification, ending with a human reviewing the evidence and making the final decision.
Rather than adding more instructions to an increasingly large prompt, you’ll learn how to structure the engineering layer around the model so your AI-assisted development workflows are easier to reuse, control, and maintain.
What You’ll Learn
By the end of the webinar, you’ll understand how to:
Build an agent harness around Claude Code and understand the role each component plays
Decide what belongs in a prompt, CLAUDE.md, Agent Skill, tool, or subagent
Turn repeated instructions and workflows into reusable Agent Skills
Use progressive disclosure to keep the active context focused
Introduce specialist subagents for tasks such as testing, verification, and review
Use hooks to automatically run deterministic checks such as formatting, linting, and tests
Combine model reasoning, tests, guardrails, and human review
Build a complete workflow from feature request → implementation → verification → human decision
See how the same patterns can extend into more advanced agentic engineering systems Free 60 min Webinar Luca + Yous…
Who Is It For?
This webinar is designed for:
Software developers and engineers
AI application engineers
Backend and full-stack developers
Platform and DevOps engineers
Technical leads and solutions architects
Claude and Claude Code users
Developers experimenting with AI coding agents who want to move beyond basic prompting
Level: Beginner to intermediate. Basic programming knowledge and some familiarity with AI coding tools or LLMs is recommended. No previous experience with Agent Skills is required.