#harnessed.md
"The primary job is no longer to write code, but to design environments, specify intent, and build feedback loops that allow agents to do reliable work."
— OpenAI, Harness Engineering
##What is a harnessed company?
Engineering has fundamentally changed.
Traditionally engineers spent 80% of their time on features, 20% on the system. Harnessed companies flip that on its head: 80% goes to the harness — the machine that builds the machine. 20% guiding where it goes.
The harness has three parts:
- Guides (feedforward) — steer the agent before it acts: AGENTS.md, design docs, architecture maps, rules, learnings
- Verify (pre-ship) — check the work before it ships: types, linters, tests, agentic review
- Observe (post-ship) — watch it in production: error monitoring, usage patterns, performance
Intent ◄··········································╮
│ new tasks + bug fixes ·
▼ ·
Guides ◄·····································╮ ·
│ · ·
│ constrain + direct evolve · ·
▼ · ·
Agent builds ◄───────┐ · ·
│ │ · ·
▼ │ no · ·
Verify ··············╁·······················┤ ·
│ │ · ·
▼ │ · ·
Pass? ───────────────┘ · ·
│ · ·
│ yes · ·
▼ · ·
Ship · ·
│ · ·
▼ · ·
Observe ·····································┴····╯
The system gets better over time. Signals from verification and observation feed back into the guides — you don't just fix the code, you evolve the harness so it never breaks that way again. That's what makes it engineering, not configuration.
##Key reading
- My AI Adoption Journey — Mitchell Hashimoto The origin of harness engineering
- Harness Engineering — Ryan Lopopolo / OpenAI 1M lines of code, zero written by humans
- Harness Engineering for Coding Agent Users — Birgitta Böckeler / ThoughtWorks The guides and sensors framework
- Compound Engineering — Dan Shipper, Kieran Klaassen / Every How Every codes with agents