Summarizes captured clippings into interlinked Obsidian wiki source and concept pages.
npx skills add https://github.com/ehartye/wiki-master --skill wiki-ingest-ehartye
Please help me install this Agent Skill. Skill: wiki-ingest Source: https://github.com/ehartye/wiki-master/tree/main/skills/wiki-ingest Command: npx skills add https://github.com/ehartye/wiki-master --skill wiki-ingest-ehartye
health.mjs --backlog) so only unprocessed clippings are handled. - Provenance-Safe Summaries: Writes wiki/sources/<slug>.md pages with quoted wikilink citations, exact clipping paths, and source-hashes to prevent broken links and provenance gaps. - Concept Management: Searches for existing concepts before creating new pages, extends canonical pages, adds aliases, and updates the task map and generated catalog. - Atomic Commits: Wraps the operation in op-begin/op-commit so each ingest is committed as one revertable unit without touching the user's uncommitted work. - Use Case: After clipping a dozen articles on a research topic, run the ingest to turn each raw clipping into a cited source summary, link them into existing concept pages, and regenerate the wiki index. ## Quick Start Ask the agent to process the wiki ingest backlog, or pass the path of a specific raw clipping to summarize it into the wiki.High-intent search queries and answers about installing and using this skill.
Run the skill with no arguments; it computes the backlog using the health.mjs --backlog hash-join metric, which lists clippings whose hashes have no matching summary. If the count is zero, nothing is pending and the run stops.
Pass the path or name of a raw clipping as the argument. The skill reads that exact file, writes or updates wiki/sources/<slug>.md with a summary, quoted wikilink citations, and the clipping's source-hash, then regenerates the catalog.
This happens when citations are retyped from a source title instead of copied from the actual clipping path, since the clipper slugifies special characters into filenames. Use repair-provenance-links.mjs with a dry run, then --apply, to fix drifted links.
Unquoted wikilinks like sources: [[A]] parse as nested YAML lists, not strings. Always quote each wikilink, e.g. sources: ["[[A]]", "[[B]]"], and run repair-inline-sources.mjs to fix existing pages in bulk.
Parallel ingests are discouraged because two concurrent runs can overwrite shared concept pages. Prefer a named clipping set and coordinate overlapping targets rather than processing the global backlog simultaneously.
It only summarizes clippings already captured in raw/; it never edits raw/ and cannot fetch new sources. Uncaptured sources require a separate discover or clip step, and it depends on the obsidian CLI and Node.js helpers.
00PZRoute Obsidian Web Clipper markdown clippings into niche knowledge bases.
paperclipaiRetrieve and synthesize information from wiki pages and raw sources with citations.
pax-kDetect orphaned pages, broken wikilinks, and frontmatter issues in Obsidian wikis.
pax-kAudit wiki state by computing ingested content and delta from sources.