recall

Recall provides persistent memory for local Code sessions, ensuring continuity and efficiency without compromising privacy or incurring additional costs. Key features include: • Local session logging • On-device session summarization • Resume-ready context generation • Operates without external APIs or cloud services • Cost-free on your existing subscription This tool addresses the challenge of Code starting each session without memory. It acts as a local logbook, incrementally capturing session activities and then condensing them into a concise, relevant summary with a classical Python summarizer based on TF-IDF and TextRank. This process is entirely local, ensuring no data ever leaves your machine, safeguarding sensitive code, paths, and potential secrets. The system integrates seamlessly into your workflow. During a session, actions are appended to `history.md`. At the start of a new session, `context.md` surfaces as a prompt to resume work. When you're ready to wrap up, a simple command triggers the local summarizer to update `context.md`, providing a "where are we right now" overview for future reference. This design eliminates friction, requires no installation of local models, and works offline. Built for developers and engineering teams using local Code subscriptions who prioritize data privacy and cost efficiency. It's ideal for maintaining context across coding sessions, facilitating project handovers, and quickly getting up to speed on previous work.