Open-source pipeline classifies phone screenshots, then analyzes and searches the web
TheMoonMidas · x · 2026-09-05
A new open-source repo (stefanionescu/screenshot-analysis-pipeline) offers a Python CLI for analyzing phone/desktop screenshots. It runs the yapwithai/phone-screen-classifier locally via ONNX Runtime to return a screen type and safety label, picks an analysis prompt based on the classification, generates research queries about the screenshot's topic, and searches the web via Exa or Gemini for context. The author optimized each step for speed, completeness, and web context, noting GPT Luna has the best speed/quality tradeoff among frontier models. All prompts ship in the repo.
More from coding & agent
- GPT-6 Astra flunks complex PCB routing after 2h20m and 15% of weekly limits in biggest public test — yacineMTB · 2026-09-05
- Fable 5.1 medium effort matches Fable 5 high, no longer breaks prompt cache — lydiahallie · 2026-09-05
- GPT-6 Astra's first task uncovers two bugs from GPT5.6 Sol fix — op7418 · 2026-09-05
- GPT-6 Astra's first task: quickly finds two bugs introduced by an earlier model's fix — op7418 · 2026-09-05
- Astra churns 2 hours via Figma MCP to rebuild an entire design system as a component library — AIandDesign · 2026-09-05
- Dev Combines Grok Bot and Hermes Agent Into Persistent Self-Improving Personal Agent — omarsar0 · 2026-09-05