topic-graph

You can discover recurring semantic topics across agent memory by constructing a time-decayed topic graph that extracts themes and ranks them by time-decayed importance, surfacing recent concepts while preserving historical patterns.

To query topic relations and hierarchies in memory, use the memory-daemon CLI with commands like 'topics query' to explore specific themes and 'topics related' to inspect connected topics and navigate hierarchical TOC nodes.

Time-decayed topic scoring is a memory analysis method that calculates topic significance by weighting recent conversations more heavily, allowing you to monitor topic trends over time while still retaining awareness of historical patterns.

The topic graph requires no external dependencies or database setup, operating entirely through the memory-daemon CLI to extract semantic topics and retrieve related hierarchies from existing agent memory without additional packages.

Yes, you can check the health and status of your semantic topic graph by running the 'memory-daemon topics status' command, which verifies that the memory analysis environment is correctly constructed and ready for querying.

The best way to monitor topic trends over time in conversations is by using the 'memory-daemon topics top' command to retrieve and rank themes by time-decayed significance, revealing how discussion clusters shift across memory.