Google Cloud engineer opens MIT-licensed harness guide for better agents without changing the model
blaizedsouza · x · 2026-07-22
What it is
A Google Cloud engineer, Ryan Lopopolo, open-sourced harness-engineering: an MIT-licensed anthology and field guide for improving agent performance without changing the model.
What the repo contains
- A structure for agent context bundles and references to put into CLAUDE.md
- Guidance around the full harness / loop around the model
- Docs on topics like authority, continuous maintenance, domain modeling, durable systems, feedback, fixed workers, just-in-time context, last-mile deployment, lineage, proof, tool legibility, and whole-job workflows
- Playbooks for upgrading a harness and reviewing a repo for agent readiness
- Source archives, notes, evals, and worked examples
Why people are sharing it
The poster argues this is unusually strong open material on harness engineering—the surrounding system design that can make agents dramatically better even when the underlying model stays the same.
More from coding & agent
- Cognition's SWE-2 uses a KKT duality argument in RL to shift the effort Pareto curve — YouJiacheng · 2026-09-11
- First-ever Three.js Conference lands in Paris, with a panel on AI-shortened design workflows — OdinLovis · 2026-09-11
- Data engineering, not agent frameworks, is the real bottleneck for enterprise AI agents — dhruv2038 · 2026-09-11
- RTK Terminal Compression Cuts Tokens but Leaves Your AI Coding Bill Unchanged — Bartaseth · 2026-09-11
- GPT-6 Astra beats Factorio with enemies in 44 in-game hours at ~$4,500 API cost — liminal_bardo · 2026-09-11
- Investment Analyst Asks How to Build a Claude-Based Diligence Agent Stack — Careless_Tie2286 · 2026-09-11