remember

You can store learnings for future recall by saving user-defined solutions and decisions into a persistent memory system using PostgreSQL and BGE embeddings, which automatically extracts tags for easier retrieval.

Specific learning types you can persist include WORKING_SOLUTION, ARCHITECTURAL_DECISION, CODEBASE_PATTERN, FAILED_APPROACH, and ERROR_FIX, managed via auto-detection or manual assignment.

Save an error fix by classifying it as an ERROR_FIX learning type and storing it in the PostgreSQL database, where automatic tag extraction ensures the knowledge is easily retrieved if the issue arises again.

Storing knowledge patterns does not require manual tagging because the system automatically extracts relevant tags from the content directly to enhance future recall alongside type classification.

You need PostgreSQL to persist architectural decisions and working solutions, as the memory system relies on it alongside BGE embeddings to store and manage your classified learnings.

Save a failed approach to a persistent memory database immediately after identifying it does not work, categorizing it as a FAILED_APPROACH to prevent repeating the same debugging efforts later.