Manual Prompt Writing Beats Automated Agent Optimization, Dev Finds
BraceSproul · x · 2026-08-03
Developer Brace Sproul shared practical insights on prompt optimization, noting that humans are still crucially needed in the loop. While agents are useful for reading traces and running scripts, manual prompt writing often yields better results.
Sproul spent about three days trying to automate the optimization of the OpenWiki prompt using Codex via hill climbing, but eval scores barely increased. Switching to full manual mode, he made more progress in a few hours than the agent did in three days.
More from coding & agent
- Developer demos what an AI agent sees while trying to break out of its sandbox — cramforce · 2026-08-03
- Kimi K3 Full Course: From Workflows to Multi-Agent Orchestration — ZabihullahAtal · 2026-08-03
- Open-Source Chrome Extension Turns Web Images into 3D Models via fal.ai — tom_doerr · 2026-08-03
- Comp AI Open-Sources Agentic CRM on Next.js, Launches 'eve' Agent Framework — JosephJacks_ · 2026-08-03
- Building Flows: An Evidence and Verification Layer for AI Coding Agents — OGMYT · 2026-08-03
- AI Coding Era Reshapes DevTools: Open Source Gains Edge Over Closed — davidcrawshaw · 2026-08-03