Show HN: ambient-context Builds Screen Memory From Text Instead of Screenshots
Hacker News·low signal
ambient-context captures what is on screen as text converted to Markdown rather than storing screenshots, then feeds that as context to agents. The design sidesteps the storage and vision-token cost that kills most screen-recall tools, and keeps the archive greppable. It hit 56 points and 22 comments on Hacker News, so this is an early project rather than a proven one.