Show HN: SlimSnap – mark a screenshot element, get JSON for your coding agent (slimsnap.ai)

SlimSnap has introduced a revolutionary tool that transforms screenshots into machine-readable JSON, solving a significant limitation for coding agents like Claude Code and Codex CLI, which currently cannot process images. Users can easily capture any part of their screen, annotate it, and convert it into JSON that these agents can utilize for programming tasks, thus bridging the gap between visual content and code. With a simple command in macOS, the tool not only reduces token usage—approximately 55% less on Sonnet and up to 85% on Opus 4.7 and 4.8—but also enhances the ability of these agents to interface with UI elements precisely, eliminating guesswork about positioning and context. The significance of SlimSnap lies in its technical efficiency and accessibility. It runs locally on macOS without requiring uploads, ensuring user privacy and security. The tool implements built-in OCR for reading UI elements, and each captured JSON schema details all aspects of the screenshot, including element types and coordinates, making it easier for agents to perform specific coding tasks. The approach is particularly beneficial for iterative development workflows where precise communication about UI components is critical, promising a more seamless integration of visual design into coding processes while fostering improvements in developer productivity and collaboration.