AgentMemory Skill

To persist agent memory across sessions, you can use a local SQLite-backed database to store facts, lessons, and entity profiles. This enables the retrieval of context during new conversations or tasks.

Preventing AI agents from repeating mistakes involves recording lessons learned from negative outcomes. You can store these experiences in a database and review negative lessons to avoid repeating errors in similar contexts.

Yes, you can track user preferences and entity attributes using a SQLite database. It supports FTS5-based semantic-style text search to retrieve tagged facts and maintain stable profiles for people or projects.

Agent memory supports automatic cleanup of stale facts and lessons through optional expiration and access-based forgetting. This removes outdated information to keep the stored knowledge relevant.

You can store reusable technical constraints as facts with tags and confidence levels. Later, retrieve this knowledge using full-text search to recall relevant context for ongoing tasks.