graph-explore

name	graph-explore
description	Explore a graph or network dataset — structure, statistics, paths, and visualization. Use when working with DOT, GML, GraphML, or Mermaid files.
allowed-tools	mcp__localdata__connect_database mcp__localdata__get_graph_stats mcp__localdata__get_edges mcp__localdata__get_neighbors mcp__localdata__get_node mcp__localdata__find_path mcp__localdata__export_graph
argument-hint	<file-path>

Graph Explore

Connect to a graph file, analyze its structure and key nodes, find paths, and export visualizations.

Steps

1. Connect to the graph. Call connect_database with the file path from $ARGUMENTS. The tool auto-detects graph formats including DOT, GML, GraphML, and Mermaid. Note the assigned database name.

2. Get graph statistics. Call get_graph_stats with the database name. Review: node count, edge count, density, whether the graph is directed or undirected, connected components count, and average degree. This gives an overview of the graph's scale and connectivity.

3. Identify hub nodes. From the stats, note nodes with the highest degree (most connections). Call get_neighbors for the top 3 highest-degree nodes to understand what they connect to. These hubs are often the most important entities in the network.

4. Explore structure. Call get_edges to retrieve a sample of edges. Look for patterns: are edges weighted? Do they have labels or types? Is the graph sparse or dense? Identify any isolated components or bridges.

5. Inspect specific nodes. For nodes of interest (hubs or user-specified), call get_node to retrieve node attributes and metadata. Note any labels, types, or properties that provide context.

6. Find paths. If there are at least two notable nodes, call find_path between them. Review the shortest path length and the intermediate nodes. This reveals how information or relationships flow through the network.

7. Export a visualization. Call export_graph with a suitable format (DOT for Graphviz rendering, or the original format for round-tripping). For large graphs, suggest filtering to a subgraph around nodes of interest before exporting.

8. Summarize findings. Present:

   • Graph type (directed/undirected, weighted/unweighted)
   • Scale: node and edge counts, density
   • Key structural features: hubs, communities, bridges
   • Notable paths or relationships discovered
   • Suggestions for further exploration or subgraph analysis