From 'when to stop' to loop engineering: agent harnesses shape model training
omarsar0 · x · 2026-09-30
Elvis Saravia traces the evolution of agent engineering: the early problem was telling the agent when to stop, then came "loop engineering." He highlights the Codex agent and harness, noting that harnesses inform how the models themselves are optimized — making harness design a core lever of agent capability.
Related event: Agent harness and loop engineering are reshaping model optimization(2 posts)→
More from coding & agent
- Vorflux partners with OpenAI: ChatGPT subscriptions now OAuth into its agent harness — ycombinator · 2026-09-30
- Open Dots: a 4.3k-star open-source, self-hosted alternative to OpenAI's Dots — matchaman11 · 2026-09-30
- ChatGPT subscription now works in 16+ partner products including Devin and Notion — charliermarsh · 2026-09-30
- Armin Ronacher: Pi's default theme now follows your terminal's color scheme — mitsuhiko · 2026-09-30
- VendorPulse adds an organizational memory layer to procurement, going beyond static supplier dashboards — GautamRaju2107 · 2026-09-30
- Indian team builds an agent that turns Instagram Reel job posts into structured listings, with memory via Hindsight — Such-Pear-588 · 2026-09-30