Memory manager

When to use

• •Reviewing what memories exist for a project or across all projects
• •Cleaning up stale, duplicate, or low-value memories
• •Bulk-importing learnings from documentation, READMEs, or past notes
• •Searching for specific patterns, decisions, or insights

Instructions

Reviewing memories

1. •Use list_memories with no filters to see all stored memories.
2. •Group results by project and category to identify coverage gaps.
3. •Flag duplicates or contradictory memories for cleanup.

Curating memories

1. •For each memory, assess: would a future session benefit from this?
2. •Remove trivial entries with forget.
3. •Consolidate duplicates into a single, well-written memory using forget + remember.
4. •Ensure every memory has descriptive tags and the correct category.

Bulk importing

1. •When the user shares a document, README, or set of notes, extract discrete learnings.
2. •Store each as a separate memory with appropriate category and tags.
3. •Use context category for project-level setup and configuration notes.
4. •Use pattern for coding conventions and architectural patterns.
5. •Use decision for choices with rationale (e.g. "chose Postgres over Mongo because...").

Searching

1. •Use recall for semantic search — describe what you need in natural language.
2. •Use list_memories with project or category filters for structured browsing.
3. •Combine both: recall to find relevant memories, then list to see neighbors.