{"wiki":{"id":103,"slug":"semantic-memory","entity_type":"app","entity_id":"semantic-memory","title":"Semantic Memory","prose_md":"## About\n\n**Semantic Memory** is Carol's \"brain filing cabinet.\" It takes the contents of memory files — the notes, logs, and records Carol's team of agents creates — and converts them into a special kind of index called a **vector search** (think of it as a smart, meaning-based filing system that understands concepts, not just keywords). When Carol needs to remember something quickly — like \"what did we decide about the cloud migration budget last week?\" — Semantic Memory lets her find the right file by meaning, not just by file name. Built by [[agt_020]] (Merlin, Head of Execution), it acts as a behind-the-scenes librarian for Carol's entire ecosystem. Because it relies on a technology called **ChromaDB** (a purpose-built database for these smart indexes), it's a specialized service that other apps and agents call upon when they need to search across their stored knowledge.\n\n## Usage Patterns\n\nSemantic Memory fires whenever an agent in Carol's system needs to retrieve a past memory based on what it *means*, not just what it's called. For example: suppose [[agt_025]] (Sage, Analyst) is writing a new report on quarterly costs. Sage can ask a question like \"find all notes from January about over-budget alerts.\" Semantic Memory will search through thousands of memory file entries — even if none of them literally contain the words \"January\" or \"over-budget\" — and return the conceptually closest matches. It's also called upon by apps like [[carol-context]] when Carol needs to quickly recall similar past decisions during a conversation. In short: any time Carol's team needs to *remember intelligently*, Semantic Memory quietly does the heavy lifting.","namesake_json":"{}","profile_pic_path":"","source_hash":"a9c9595b1f5522ed48e55161c54d6ba5a5750c20383dc80bac4c43f31ddaed46","status":"active","last_generated_at":"2026-07-01 03:25:10","created_at":"2026-07-01 03:25:10","updated_at":"2026-07-01 03:25:10"},"facts":{"id":"semantic-memory","name":"Semantic Memory","port":7141,"url":"","description":"Vector search service for memory files (requires chromadb)","owner":"agt_020","dir_name":"semantic_memory","card_group":"","is_public":0,"access_level":"auth","log_name":"semantic-memory","gen_nginx":1,"user_admins":"[]"}}