neural-memory

You can retain and recall context across sessions by using persistent associative memory that stores decisions, errors, and causal history in a local SQLite database. This approach enables cross-conversation recall and injects recent context into new sessions automatically.

Associative recall works by using graph traversal and spreading activation to find related memories instead of simple keyword matching. This mechanism applies Hebbian learning to strengthen connections, ensuring that retrieving past decisions traces relevant contextual relationships accurately.

Yes, you can detect conflicting memories and trace causal history over time using built-in contradiction detection and temporal reasoning. This allows you to trace why something happened, what led to it, and how events relate when a project changes direction.

No, neural memory does not require external data transmission because it relies on a local SQLite-backed brain storage system. This ensures persistent associative recall and context retrieval remain entirely private without sending your data to external servers.

The best way to start a new conversation with relevant past context is to ask the assistant to recall the most relevant past decisions and context for the current task. This triggers automatic capture and context injection for immediate workflow pattern detection.

Yes, you can integrate persistent memory with MCP for workflow pattern detection through optional MCP integration. This allows you to retrieve facts, decisions, and errors while applying temporal reasoning and associative search within your existing model context protocol setup.