Is the topic graph shared across different projects?

The topic graph is per-project. This ensures that the extracted themes and relationships are specific and relevant to the codebase and conversations of the current project.

How does the topic importance scoring work?

It uses an exponential time decay formula with a 30-day half-life. Recent topics receive full weight, while topics from 30 days ago receive 50% weight, ensuring the most relevant current themes surface first.

What is the difference between active and dormant topics?

Active topics have a high importance score based on recent mentions. Dormant topics are older themes that haven't been discussed recently but remain in the system to be resurrected if they are mentioned again.

Can I use Topic Graph for keyword matching?

No, for exact keyword or phrase matching, you should use a BM25-based search skill. Topic Graph is designed for semantic relationships and conceptual exploration.

Does this skill require any background services?

Yes, it requires the memory-daemon to be running to process extractions and maintain the topic graph health.

Topic Graph Memory Explorer

Name: Topic Graph Memory Explorer
Author: SpillwaveSolutions

bySpillwaveSolutions

•

数据科学与机器学习

Extracts and visualizes semantic topic patterns from agent conversation history using time-decayed importance scoring.

The Topic Graph skill provides a sophisticated semantic layer for the agent-memory system, allowing users and AI agents to navigate conversation history through conceptual themes rather than simple keyword matches. By leveraging time-decay algorithms, it prioritizes currently relevant topics while maintaining a searchable hierarchy of historical context. This skill is essential for developers who need to track project evolution, identify recurring architectural patterns, or quickly surface related discussions across long-running development sessions within the Claude Code environment.

主要功能

01Semantic topic extraction and clustering from conversation segments

02Topic-guided navigation to specific memory nodes and segments

03Relational mapping including parent-child and similar topic connections

04Lifecycle management for active, dormant, and resurrected topics

05Time-decayed importance scoring to prioritize recent and frequent themes

064 GitHub stars

使用场景

01Discovering recurring architectural patterns and technical themes in a project

02Mapping connections between distinct technical concepts discussed over long timeframes

03Navigating large conversation histories using topic-based semantic search instead of keywords

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add spillwavesolutions/agent-memory topic-graph

For use in Claude.ai and ChatGPT

Download Skill