Semantic search in 📝MythOS finds memos by meaning rather than keyword match. Ask a question in natural language and the system retrieves the most relevant content from your library using AI-powered vector embeddings and hybrid scoring.
How It Works
Every memo is split into chunks, enriched with its title and tags, and embedded using OpenAI's text-embedding-3-small model. When you search, your query is embedded the same way and compared against every chunk in the library. A hybrid scoring system combines vector similarity (meaning) with text matching (keywords) to surface the most relevant memos, ranked by a composite score.
Semantic search respects MythOS's permission model. You see your private memos. Collaborators see shared content. Visitors see only what's public.
Key Capabilities
- Meaning-based retrieval — finds memos about a concept even when the exact words don't match. "How does authentication work" surfaces auth memos regardless of terminology
- Hybrid scoring — combines 70% vector similarity with 30% text relevance, catching both conceptual matches and keyword hits
- Chunk-level precision — results include the specific text passages that matched, so you can jump straight to the relevant section
- Permission-aware — filters are applied at the database level, not post-retrieval. Private content is invisible to non-owners
- Graceful degradation — when vector search is unavailable, falls back to text-only results with a diagnostic flag
Getting Started
Connect your library to an AI client via 📝MCP, then ask it to search semantically. See 📝How to Use Semantic Search in MythOS for step-by-step instructions.
Related
- 📝Search (MythOS) — parent feature overview covering keyword and semantic search
- 📝Semantic Search API — technical documentation for the MCP tool and REST endpoint
- 📝How to Use Semantic Search in MythOS — end-user tutorial
- 📝How to Chat with Your Library via MCP — RAG-powered chat uses the same retrieval pipeline
- 📝MCP vs RAG — the distinction between connectivity and retrieval
Thoughts
Semantic search was the missing piece between having a knowledge library and having the library actually be useful to AI agents. Keyword search finds what you remember naming. Semantic search finds what you forgot you wrote. The difference is the difference between a filing cabinet and a mind.