LLM Wiki Lint

SkillDocs & knowledge

Validate Wiki and source note health. Use before commits, after ingests, or when notes were edited by hand.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the LLM Wiki Lint skill

What this skill tells your AI

The instructions your AI receives, as published by gavishap/omnia-vault in .claude/skills/llm-wiki-lint/SKILL.md and read by ahel’s review.

Run the deterministic checks and fix what they report.

Steps

  1. Run the full check suite:

    python scripts/wiki_tool.py doctor
    python scripts/wiki_tool.py build
    python scripts/wiki_tool.py lint
    python scripts/wiki_tool.py source-lint
    python scripts/audit_public.py
    
  2. For every lint failure, open the offending note and fix the root cause:

    • wrong or missing tag → set exactly one allowed tag matching the folder
    • source_count mismatch → recount sources entries
    • broken source link → fix the path or restore the missing Raw source
    • missing source frontmatter → fill in Title, Reference, Created, Processed, tags
    • processed-but-uncovered source → either compile it into Wiki notes or set Processed: false
  3. Re-run until everything passes.

  4. See Schema/lint-checklist.md for the full rule list, including the manual checks lint cannot automate (claim support, no invented citations).

Rules

  • Fix causes, not symptoms: do not silence a check by deleting sources or downgrading a note's frontmatter.
  • Never hand-edit generated files (Wiki/catalog.jsonl, index files, Schema/source-manifest.jsonl); regenerate them instead.

Signals

GitHub stars
62
Forks
6
Last commit
Aug 2026
Advanced
Catalog kind
skill
Gateway key
llm-wiki-lint
Source
github.com/gavishap/omnia-vault