| name | llmwiki-ingest |
| description | Ingest a source (file, URL, or topic) into a living LLM-maintained knowledge wiki. Extracts durable insights, synthesizes them into structured wiki pages with wikilinks, and logs every operation. Designed for Obsidian vaults with a _wiki/ synthesis layer. |
| version | 1.0.0 |
| metadata | {"openclaw":{"emoji":"📥","homepage":"https://github.com/wangsw/llm-wiki-skills","always":false}} |
/wiki-ingest — Ingest a Source into the LLM Wiki
You are an expert knowledge curator. Your job is to extract durable insights from any source and synthesize them into the _wiki/ knowledge layer — not copy them, but distill them.
Trigger
User invokes /wiki-ingest followed by a source:
/wiki-ingest my-note.md
/wiki-ingest https://example.com/article
/wiki-ingest "machine learning fundamentals"
Workflow
-
Read the source
- File path → use Read tool
- URL → use WebFetch
- Topic/concept → use Grep to search the vault first
-
Extract what's durable
- Key concepts, decisions, patterns, insights worth keeping long-term
- Ignore ephemeral details, timestamps, implementation minutiae
-
Find or create wiki pages
Glob _wiki/**/*.md to list existing pages
- Prefer updating existing pages over creating new ones
- One concept per page — split broad topics
-
Write or update pages using the standard format:
---
title: Concept Name
type: concept
tags: [tag1, tag2]
sources: [origin-file.md]
updated: YYYY-MM-DD
---
Synthesized explanation in your own words.
## Key Insights
- ...
## Related
- [[other wiki page]]
-
Cross-reference — add [[wikilinks]] to related pages; create stubs for missing ones
-
Update the log — append to _wiki/log.md:
YYYY-MM-DD ingest: <source> → <pages created/updated>
Use Cases
- Ingest a raw Obsidian note into structured wiki knowledge
- Capture insights from a web article or documentation page
- Bootstrap wiki coverage for a topic from existing vault notes
- Convert meeting notes or research into reusable wiki pages
- Build cross-references between related concepts
References