Library Search is how you find memos inside your own 📝MythOS Search library — two complementary modes that cover the two ways you remember things. Keyword search finds a memo by the words in its title or tags; semantic search finds it by what it's about, even when you can't recall the exact wording. Together they serve both the organized mind and the associative one.
How It Works
Keyword search matches a case-insensitive substring against memo titles and tags. Type a word to surface every memo whose title or tags contain it; prefix the query with # to match a tag exactly, which turns any tag into a browsable collection. It runs everywhere the feed appears — the creator page, the memo feed, and the command palette — and always respects visibility, so a visitor only ever sees what you've made visible to them.
Semantic search goes by meaning rather than letters. Every memo is chunked, enriched with its title and tags, and embedded as a vector; your query is embedded the same way and compared across the library, with a hybrid score blending meaning and keyword overlap. This mode powers AI retrieval — the search_memos tool and the RAG pipeline behind chatting with a library — rather than the in-app search bar.
Features
- 📝Keyword Search — substring matching over titles and tags, with
#for exact tags - 📝Semantic Search — meaning-based retrieval over your library's content via embeddings
Getting Started
Use the search box on your library feed and type a title word — or a #tag for an exact collection. For semantic search, connect your library to an AI client via 📝MythOS MCP and ask it to search.