Weekly Reflect

To consolidate agent activity into atomic notes, you run an automated reflection cycle that processes recent memory logs, extracts durable claims, and saves them as individual notes under the memory directory. This ensures long-term memory remains structured and easily navigable.

A memory-health check evaluates the cleanliness and navigability of your note graph by identifying stale entries and structural drift. You need to run a weekly reflection cycle when recent agent activity accumulates and requires pruning to maintain efficient recall.

Generating MOCs from atomic notes involves the Skill scanning recently consolidated memory entries to identify durable topics, then automatically creating Map of Content files that link related atomic notes together. This updates the notegraph delta for review and sustains navigability.

Yes, you can prune stale memory entries during the weekly reflection cycle. The Skill identifies outdated information, removes it from the memory graph, and generates a notegraph delta log that records all structural changes for your review.

The best way to organize durable claims is applying atomic-notes conventions, where each note contains a single idea with standardized frontmatter metadata. The Skill automates this by extracting claims from recent activity and structuring them as reproducible graph nodes.

No dependencies are required to manage memory notes. The Skill operates independently by reading recent agent activity, applying atomicity rules, and writing consolidated notes with conforming frontmatter directly into your existing memory directory structure.