One Prompt Turns Gemini Flash Into an Optimization Machine: 5000x Gains in 20 Minutes
doodlestein · x · 2026-08-29
The author shares an agent coding workflow: have the model carefully read AGENTS.md and README.md first, use code investigation mode to fully understand the architecture, then chain /profiling-software-performance with /repeatedly-apply-skill (at least 30 iterations) to apply /extreme-software-optimization.
On the FrankenMarkdown project, this produced several 5000x+ performance gains in 20 minutes on code written by other models, plus multiple 2x+ improvements. All done on Gemini 3.7 Flash, which costs essentially nothing relative to the value of your time.
More from coding & agent
- PRAXIST coordinates AI agents for complex tasks like rocket landing — iamfakhrealam · 2026-08-29
- fbtee 4.0 released, fully rewritten in Rust with Oxc — cnakazawa · 2026-08-29
- OpenAI cuts off Cursor, Anthropic cuts off Windsurf, revealing vendor lock-in risks — Yuchenj_UW · 2026-08-29
- Together AI: cascading GLM-5.3 Flash to GLM-5.3 cuts cost 57% while boosting DeepSWE to 80.9% — togethercompute · 2026-08-29
- Google Paper: Replace Agent Chat History With Explicit State, Cut Tokens 16x — rohanpaul_ai · 2026-08-29
- Seeking open-source methods to extract tables from Indian bank PDFs — OmPatel110 · 2026-08-29