Currently building SnipAgent, a desktop snipping tool that combines screenshots with AI. After taking a snip, it uses OCR to read text (and now images), then lets users run prompts through either a local LLM (currently Qwen 1.5 Coder because thats what my machine can currently handle locally) or their own Claude, Gemini, or OpenAI API keys. Workflow: Snip → Action Menu with preset/custom prompts → AI-generated response.