memory-system

You can persist context across coding sessions by recording decisions, bugfixes, and patterns as structured observations in SQLite. This enables full-text searching of past learnings, ensuring knowledge compounds over time without redundant work.

Progressive disclosure fetches compact search results first, allowing you to retrieve full details only for relevant items. This mechanism prevents context bloat by minimizing the amount of loaded memory data during development sessions.

Document session handoffs by recording the completion state, current blockers, and next steps. This facilitates seamless transitions between development sessions by providing immediate context for continued coding tasks.

Yes, SQLite with FTS5 provides efficient full-text searching specifically for development observations. It indexes structured learnings like bugfixes and patterns, allowing you to quickly retrieve past solutions using search queries.

The best way to record bugfixes is by storing them as structured observations within a persistent memory system. This allows future search queries to instantly retrieve past solutions, preventing redundant debugging efforts.

Search memory before starting a new coding task to find similar past problems or solutions. This avoids redundant work by loading relevant prior observations, ensuring you build upon existing knowledge rather than repeating previous efforts.