wiki-ingest

Ingests raw sources into an Obsidian wiki vault with classification, deduplication, and archiving.

Updated May 10, 2026
One-click install
npx skills add https://github.com/cookkie03/skills --skill wiki-ingest-cookkie03
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: wiki-ingest
Source: https://github.com/cookkie03/skills/tree/main/wiki-ingest
Command: npx skills add https://github.com/cookkie03/skills --skill wiki-ingest-cookkie03

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Keeping a personal knowledge wiki up to date requires manually reading raw notes, articles, and transcripts, deciding where each piece belongs, and avoiding duplicates. This Skill automates that ingestion pipeline for an LLM-powered Obsidian vault. ## Core Features & Use Cases - Autonomous raw folder processing: Reads everything in raw/ (excluding raw/archived/), classifies each file as new content, an update, or user instructions, and builds an execution plan. - Anti-duplication and conflict handling: Checks _meta/index.md for existing pages by title or raw_source_path, updates instead of duplicating, and applies a conflict policy based on updated dates and confidence levels. - Structured closure: Archives processed files, updates the index and hot-cache, and appends ingest/update entries to a log. - Use Case: Drop a meeting transcript and a saved article into raw/, then let the Skill create source pages, update related entity and knowledge pages, and archive the originals. ## Quick Start Ask the assistant to ingest everything currently in the raw folder of your wiki vault and update the related pages.

Frequently Asked Questions about wiki-ingest

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I ingest new notes into an Obsidian wiki vault automatically?

Place the files in the vault's raw/ folder and invoke the wiki-ingest skill. It reads every file except raw/archived/, classifies the content, creates or updates wiki pages according to _meta/taxonomy.md, then archives the processed files.

How does the skill avoid creating duplicate wiki pages?

It consults _meta/index.md and searches for existing source pages with the same raw_source_path or similar titles, plus existing entity or concept pages. Matches are updated in place instead of creating duplicates, and draft pages on the same topic are completed rather than forked.

Can the skill handle audio files and other non-text sources?

Yes. For audio it looks for a declared transcription file next to the audio and uses it if present; otherwise it delegates to the wiki-preprocess skill. Images are described if the model has vision, and unsupported formats are routed through preprocessing rather than forced.

What happens when new content contradicts an existing wiki page?

The source with the more recent updated date in its frontmatter wins by default. If the existing page has confidence: high, the conflict is flagged before overwriting, and every resolved contradiction is recorded in the log during closure.

Does the skill modify the wiki without user confirmation?

Pure content ingestion proceeds autonomously, but if the plan includes executing user instructions found in raw/ (such as deletions or edits), the skill must present the plan and ask for confirmation before applying any changes.