Skip to content
Research Insights Made Simple
Engineering article reviewResearch Insights Made Simple

Harness engineering: agent environments

Five months in which Codex wrote all of a product’s code while people built its environment

/ Research Insights Made Simple · Agent harnesses 3/7 · Harness engineering

Slide contents

  1. 1. Harness engineering: agent environments

    Five months in which Codex wrote all of a product’s code while people built its environment

  2. 2. Five months without hand-written code

  3. 3. Humans steer; agents write everything

  4. 4. Agent failures signal a missing capability

  5. 5. Human QA became the bottleneck

  6. 6. Every change boots its own app

  7. 7. Codex queries logs, metrics and traces

  8. 8. What Codex can’t see doesn’t exist

  9. 9. AGENTS.md is a table of contents

  10. 10. Repo knowledge is versioned and machine-checked

  11. 11. Domain layers are enforced mechanically

  12. 12. Lint errors tell the agent the fix

  13. 13. Enforce invariants, not implementations

  14. 14. One prompt: from bug to merge

  15. 15. At high throughput, fixing beats waiting

  16. 16. Pay down debt daily, not on Fridays

  17. 17. The numbers show scale, not cause

  18. 18. Five months observed; years remain open

  19. 19. Measure each environment capability you add

  20. 20. Count accepted tasks and human time

  21. 21. Discipline moves from code to scaffolding