wiki-graph

Generate a knowledge graph of the wiki — nodes are pages, edges are [[wikilinks]].

Usage: /wiki-graph [format]

$ARGUMENTS is one of: json, html, or both (default: both).

The graph tool walks every file under wiki/ looking for [[wikilink]] references, builds a node-and-edge list, and writes:

• graph/graph.json — canonical data: {nodes: [...], edges: [...]}
• graph/graph.html — interactive vis.js visualisation you can open in a browser

Steps

1. Ensure the Python graph builder exists at llmwiki/graph.py. If it doesn't, create it (see "Fallback" below for a pure-agent version).

2. Run:

   python3 -m llmwiki graph $ARGUMENTS
   

3. Read graph/graph.json after it runs and report to the user:

   • Total node count (pages)
   • Total edge count (wikilinks)
   • Top 5 most-linked pages (entities or concepts with the highest in-degree)
   • Top 5 most-linking pages (sources with the highest out-degree)
   • Any orphan nodes (zero inbound edges)
   • Any broken edges (links to pages that don't exist) — cross-check with /wiki-lint

4. If the user asked for html or both, offer to open graph/graph.html locally with python3 -m llmwiki serve --dir graph --port 8766.

5. Append to wiki/log.md:

   ## [YYYY-MM-DD] graph | <N> nodes, <M> edges
   

Fallback (no graph.py)

If llmwiki/graph.py doesn't exist yet, build the graph manually using Grep and Read:

1. grep -roh '\[\[[^]]*\]\]' wiki/ | sort -u → unique wikilink targets
2. For each wiki page, list the wikilinks found in it (out-edges)
3. Compute the reverse index to find in-edges
4. Write graph/graph.json with the format:

   {
     "nodes": [{"id": "slug", "label": "Display Name", "type": "source|entity|concept", "in_degree": N, "out_degree": M}],
     "edges": [{"source": "from-slug", "target": "to-slug"}]
   }
   

5. Write graph/graph.html — a single HTML file with an inline vis.js embed that reads graph.json via fetch().

Report the same statistics as above.