ProjectsBuild. Measure. Iterate.

Hands-on practice comparing baseline workflows against agentic pipelines — with real code, real failures, and real improvements.

⚡

Project 01 · Vibe Coding vs Harness Engineering

Run the same feature implementation twice — first in vibe-coding mode with a vanilla agent, then with a minimal harness. Measure the difference in reliability, scope adherence, and time-to-verify.

Start Project 01

🏗️

Project 02 · Building a Minimal Harness

Scaffold AGENTS.md, feature_list.json, and progress.md from scratch for a real repository. Wire up the verification pipeline and run your first fully governed agent session.

Start Project 02

🤝

Project 03 · Multi-Agent Pipeline

Implement an orchestrator-worker pattern for a feature that spans frontend, backend, and tests. Observe coordination failures and apply the shared-state patterns from Lecture 05.

Start Project 03