wiki-stats

Wiki Stats

When to use

Use this skill when the user asks how large the wiki is, how connected it is, what changed recently, or what to work on next. It only reads files and never edits the wiki. It needs a folder of markdown pages; log.md, raw/, and frontmatter are used when present and skipped when absent.

Steps

1. Locate the wiki root. Use wiki/ if it exists, otherwise the folder the user named. Set WIKI to that path. If a sibling raw/ folder exists, set RAW to it.

2. Count pages by type. Prefer the frontmatter type and fall back to the parent folder:

   for f in $(find "$WIKI" -name '*.md' -not -name 'index.md' -not -name 'log.md' -not -name 'overview.md'); do
     t=$(sed -n '/^---$/,/^---$/p' "$f" | awk -F': *' '$1=="type"{print $2; exit}')
     echo "${t:-$(basename "$(dirname "$f")")}"
   done | sort | uniq -c | sort -rn
   

3. Measure link density. Count wikilinks per page and overall:

   total=$(grep -rhoE '\[\[[^]]+\]\]' "$WIKI" --include='*.md' | wc -l | tr -d ' ')
   pages=$(find "$WIKI" -name '*.md' -not -name 'index.md' -not -name 'log.md' | wc -l | tr -d ' ')
   echo "links=$total pages=$pages avg=$(awk "BEGIN{printf \"%.1f\", $total/($pages?$pages:1)}")"
   

   If the wiki uses standard markdown links, count ](...*.md) instead.

4. Find hubs and orphans. Rank pages by inbound links:

   for f in $(find "$WIKI" -name '*.md' -not -name 'index.md' -not -name 'log.md'); do
     stem=$(basename "$f" .md)
     n=$(grep -rlF "[[$stem" "$WIKI" --include='*.md' | grep -v "index.md" | grep -vc "/$(basename "$f")$")
     printf '%s\t%s\n' "$n" "$stem"
   done | sort -rn > /tmp/wiki-inbound.tsv
   head -5 /tmp/wiki-inbound.tsv          # hubs
   awk -F'\t' '$1==0' /tmp/wiki-inbound.tsv  # orphans
   

5. Summarize recent activity from log.md if it exists. Entries follow ## [ISO-timestamp] operation | title:

   grep -E '^## \[' "$WIKI/log.md" | sed -E 's/^## \[[^]]+\] ([a-z-]+).*/\1/' | sort | uniq -c | sort -rn
   grep -E '^## \[' "$WIKI/log.md" | tail -5
   

   Report operations by type, the five most recent entries, and days since the last ingest entry.

6. Check source coverage when RAW is set. Every raw file should have a summary page:

   find "$RAW" -type f -not -path '*/assets/*' | wc -l
   find "$WIKI/sources" -name '*.md' 2>/dev/null | wc -l
   

   Compute coverage as summaries divided by raw files. List raw files whose stem does not appear anywhere under $WIKI/sources/.

7. Present the dashboard with these sections in order: Page Counts, Link Health, Top Hubs, Orphans, Recent Activity, Source Coverage, Suggested Next Steps. Keep each section short. Base the suggestions on the numbers: un-ingested sources suggest ingesting them, orphans suggest cross-referencing or pruning, a long gap since the last ingest suggests finding new sources, and low link density suggests running lint-wiki.

What good output looks like

Wiki stats for wiki/

Page Counts     42 pages — 18 entities, 15 concepts, 7 sources, 2 comparisons
Link Health     203 wikilinks, 4.8 per page, 0 broken
Top Hubs        acme-corp (14), retrieval-latency (9), vector-databases (8)
Orphans         2 — old-pricing-model, meeting-2025-03-notes
Recent Activity 31 ingest, 12 query, 4 lint; last ingest 3 days ago
Source Coverage 7 of 9 raw files summarized (78%) — missing: q3-report.pdf, podcast-ep-12.md

Next: ingest the 2 uncovered sources, then run lint-wiki to link the 2 orphans.

When a command fails

• grep: log.md: No such file: there is no log. Skip the Recent Activity section and say so.
• find: raw: No such file: no raw folder. Skip Source Coverage and say so.
• Division shows avg=0.0 with a nonzero page count: the wiki uses markdown links, not wikilinks. Recount with grep -rhoE '\]\([^)]+\.md\)'.
• Filenames with spaces break the for f in $(find ...) loops. Rerun with find "$WIKI" ... -print0 | while IFS= read -r -d '' f; do ...; done.