v5.0.0

⚠️ Breaking change

scraps search and the MCP search_scraps tool search what names a scrap, not its body. They match a scrap's key, title, frontmatter description and tags. A word that appears only in a scrap's body no longer finds it.

There is no switch back to body search. To find text inside bodies, grep the wiki's *.md files. To keep a scrap findable by a term, put the term in its description or a tag. Read a hit's body with scraps get or get_scrap, as before.

Why: fuzzy matching across a whole body matched nearly any keyword. A search that left num unset could fill the default cap of 100 hits. It also meant every body had to be loaded for every search. The static site's search has only ever indexed titles.

What's Changed

✨ Features

• feat(search)!: search what names a scrap, not its body (#858)

  The mcp-server plugin goes to 4.0.0. The pkm learn skill and llm-wiki's ingest find plain-text mentions of a new page with Grep, since search no longer reads bodies.

🔧 Maintenance

• v5.0.0 (#859)

Full Changelog: v4.0.0...v5.0.0