Managing Memory

Unified Workspace Memory Management

This skill covers the tools, scripts, and processes used to synchronize, index, clean, and reflect upon episodic, semantic, procedural, and relational memories across the workspace and vector/graph databases.

When to Use

Use this skill when performing memory maintenance, indexing new entities, reflecting on logs, running experience collectors, or checking memory health.

Prerequisites

• Conda environment: cursor-factory python interpreter.
• Qdrant Vector DB or Memory MCP server connection active.

Process

Follow these procedures to manage the workspace memory systems.

Checking & Initializing memory

To verify memory server availability and collections:

conda run -p D:\Anaconda\envs\cursor-factory python scripts/memory/check_memory.py
conda run -p D:\Anaconda\envs\cursor-factory python scripts/memory/init_collections.py

Episodic Logging & Reflection

To log sessions and invoke the reflection engine:

conda run -p D:\Anaconda\envs\cursor-factory python scripts/memory/hook_session_start.py
conda run -p D:\Anaconda\envs\cursor-factory python scripts/memory/reflection_engine.py

Best Practices

• Isolation: Ensure memory operations run within correct database boundaries.
• Safety: Do not mutate verified memory nodes without proper approval gates.