remember

To save project context as explicit memories, provide a situation, subject/predicate, or memory note to classify it as feedback or fact, then store it with user approval in a persistent memory store. This workflow turns conversation details into structured, recallable knowledge.

Feedback represents behavioral corrections, while facts represent static knowledge. The memory capture workflow classifies your input into these two categories to apply a quality gate, ensuring the assistant saves relevant context and avoids duplicates or irrelevant data.

You can capture corrections by triggering the memory workflow with a user-provided situation or memory note. The system requests user approval, then saves the correction as feedback into a persistent memory store managed by engram.

Yes, capturing explicit memories requires user approval. The workflow uses a quality gate to classify the input as feedback or fact and prompts for confirmation before saving the structured memory to the persistent store.

Yes, you can save recall triggers by providing the specific trigger as a memory note. The system evaluates the input, determines the appropriate memory store home, and saves it with user consent to improve future assistant responses.

The memory store requires explicit user-provided situations or notes to capture knowledge; it does not autonomously extract memories. It relies on user approval and avoids saving duplicates or irrelevant data by strictly gating the classification and storage process.