What problem does it solve?
Grouping large collections of code files or embeddings by semantic similarity is difficult to do manually. This Skill runs graph community detection over a code graph so you can discover natural groupings, identify outliers, and organize vector namespaces without writing clustering code yourself.
Core Features & Use Cases
- Graph-based clustering: Runs
npx [email protected] hooks graph-cluster with spectral or Louvain community detection over code files.
- Graph min-cut analysis: Supports
hooks graph-mincut for partitioning code graphs into cohesive segments.
- Result persistence: Stores cluster assignments in ruflo memory under the
vector-clusters namespace for later retrieval.
- Use Case: Given a repository of source files, run graph clustering to reveal which files form tight semantic communities, then store the assignments to memory so agents can route related work to the same context.
Quick Start
Ask the agent to cluster the files in your project with ruvector graph-cluster and store the resulting community assignments in memory.
