Menü
Wiki knowledge base maintenance — health checks, build knowledge graph, view stats. Use for wiki integrity and graph operations. For ingest use /wiki-ingest, for queries use /wiki-query.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Basierend auf der SOC-Berufsklassifikation
| name | maintainer |
| description | Wiki knowledge base maintenance — health checks, build knowledge graph, view stats. Use for wiki integrity and graph operations. For ingest use /wiki-ingest, for queries use /wiki-query. |
You are the wiki maintainer. Follow these workflows exactly when performing wiki operations.
| Command | Skill | What to say |
|---|---|---|
/wiki-ingest | /wiki-ingest | ingest raw/my-article.md |
/wiki-query | /wiki-query | what do we know about X? |
/wiki-health | this skill | health (fast, every session) |
/wiki-graph | this skill | build the knowledge graph |
/wiki-stats | this skill | wiki stats |
Triggered by: "health" or /wiki-health
Run: python tools/health.py (or python tools/health.py --json for machine-readable output)
Fast structural integrity checks, safe to run every session:
wiki/index.md entries vs actual files on diskingest entry in wiki/log.mdOutput a health report. Use --save to write to wiki/health-report.md.
Log the operation:
python tools/log_write.py --op health --title "Health check: X issues found"
Triggered by: "wiki stats", "how many pages?", "what's in the wiki?"
Run: python tools/wiki_stats.py (or --json for machine-readable)
Fast KB inventory. Shows page counts by type and project, link density, orphan count, and last operation dates.
Triggered by: "build the knowledge graph" or /wiki-graph
Two scripts, separated by responsibility:
python tools/build_graph.py — analysis (deterministic)
[[wikilinks]] from wiki pages → edgesgraph/graph.json--report (markdown report), --save (save to graph/graph-report.md), --json (JSON report for other tools)python tools/print_graph.py — visualization
graph/graph.json and generates graph/graph.html (interactive vis.js)--open (open in browser)Typical flow: run build_graph.py first, then print_graph.py if the user wants the visual.
python tools/build_graph.py --tag tag1,tag2 — filtered subgraph
Every entry in wiki/log.md follows a parseable format:
## [YYYY-MM-DD] <op> | <title>[ | <tag>][ | <detail>]
| Field | Meaning | Example |
|---|---|---|
YYYY-MM-DD | Date of operation (ISO 8601, auto-generated) | 2026-04-25 |
op | Operation type — what happened | ingest |
title | Brief human-readable description | RAG article ingested |
tag | Project tag involved (optional) | project-alpha |
detail | Additional operation-specific info (optional) | affinity: 87% |
op | Meaning | When | Writer |
|---|---|---|---|
ingest | New document ingested | After ingest completes | log_write.py (agent) |
heal | Batch of problems fixed | After heal.py --done | heal.py (automatic) |
graph | Knowledge graph rebuilt | After build_graph.py | build_graph.py (automatic) |
report | Graph analysis report generated | After build_graph.py --report | build_graph.py (automatic) |
health | Structural health check | After health.py | log_write.py (agent) |
forget | Information removed from wiki | After forget workflow | log_write.py (agent) |
setup | Wiki infrastructure modified | Structural changes | log_write.py (agent) |
convert | File converted to markdown | After standalone /convert | log_write.py (agent) |
anonymize | Document anonymized via privacy filter | After privacy filter processes a file | log_write.py (agent) |
| Writer | Operations | How |
|---|---|---|
log_write.py | ingest, health, forget, setup, convert, anonymize | python tools/log_write.py --op <op> --title <title> |
heal.py --done | heal | Automatic in mark_done() |
build_graph.py | graph, report | Automatic in build_graph() |
| Agent direct | NONE | The agent must NEVER write to log.md directly |
grep "^## \[" wiki/log.md # all entries
grep "^## \[" wiki/log.md | grep ingest # ingest only
python tools/log_report.py --json # structured report
Remove information from the wiki knowledge base — forget projects, entities, or concepts. Use when the user wants to delete or remove wiki content.
Heal the wiki knowledge base — fix orphans, broken links, missing tags, empty pages, index desyncs. Use when the user wants to repair or clean up the wiki.
Install and verify Python dependencies for JanusLM wiki tools. Invoked automatically when the UserPromptSubmit hook detects missing dependencies, or manually via /setup.
Ingest source documents into the wiki knowledge base. Full pipeline from queue management to validation. Use when the user wants to add new documents to the KB.
Query the wiki knowledge base and synthesize an answer. Use when the user asks about topics, projects, people, or concepts that could be in the KB.
Convert non-markdown files to markdown for wiki ingest. Use when the ingest workflow encounters a file that is not .md (docx, xlsx, pptx, pdf, txt, csv, json, html, xml, etc.).