The Harness Engineering Workshop
Released 8/2026
By Vizuara
MP4 | Video: h264, 1280×720 | Audio: AAC, 44.1 KHz, 2 Ch
Genre: eLearning | Language: English | Duration: 5 Lessons ( 10h 23m ) | Size: 1.7 GB
Claude Code is not a model. It is a harness: the loop, the tools, the context machinery and the recovery systems wrapped around a model
Claude Code is not a model. It is a harness: the loop, the tools, the context machinery and the recovery systems wrapped around a model. In five mornings you build that layer from scratch and leave with your own working coding-agent harness, in the spirit of pi (pi.dev), studying Claude Code and Hermes (Nous Research) as production case studies.
Your instructor. Dr. Raj Dandekar (PhD, MIT; co-founder, Vizuara), author of Manning’s Build a DeepSeek Model (From Scratch) and creator of the Building LLMs from Scratch YouTube series.
What you build, morning by morning
Day 1 (Mon Aug 3), The Anatomy of a Harness: the agent loop from first principles; prompt vs. context vs. harness engineering.
Day 2 (Tue Aug 4), Tools and the Execution Environment: tool contracts, permission gates, sandboxing.
Day 3 (Wed Aug 5), Context Engineering Inside the Harness: context budgets, compaction, memory, the CLAUDE.md pattern.
Day 4 (Thu Aug 6), Durability, Recovery and Orchestration: checkpointing, resumable sessions, sub-agents, human-in-the-loop.
Day 5 (Fri Aug 7), Production Harnesses and Capstone: pi, Hermes and Claude Code internals; demo your own harness.
Includes daily recordings, code checkpoints for every module, and live review of every participant’s harness.
Prerequisites: comfort with Python or TypeScript and basic LLM API calls.
Questions? Write to team@vizuara.com. Workshop site: harnessengineering.vizuara.ai
HRNSENGATRFRAEERNTHAWAKKESHPPRE
you must be registered member to see linkes Register Now