Store and retrieve text as 384-dimensional semantic vectors for similarity search.
npx skills add https://github.com/jcolano/loopColony --skill zeppelin
Please help me install this Agent Skill. Skill: zeppelin Source: https://github.com/jcolano/loopColony/tree/main/TO-DO/X_DONE/X_DONE_loopcore_zeppelin_upgrade/zeppelin Command: npx skills add https://github.com/jcolano/loopColony --skill zeppelin
Provides a hosted vector search engine that removes the friction of storing, indexing, and retrieving text by semantic similarity or BM25 full-text ranking so users can build fast, production-ready semantic search over documents.
Create a 384-dimension namespace, embed your texts with the Embed API, upsert the returned vectors into that namespace, and run a similarity or BM25 query to retrieve top results.
High-intent search queries and answers about installing and using this skill.
Store and retrieve text by encoding it into 384-dimensional semantic vectors via an embedding API, then upserting them into isolated namespaces to run nearest-neighbor or BM25 queries for fast document retrieval.
Semantic vector search uses 384-dimensional embeddings to match meaning, while BM25 full-text ranking matches keyword relevance. You can run either query type against your stored documents to surface the most relevant items.
Yes, S3-native manifests handle segment compaction and provide CAS safeguards. This enables strong versus eventual consistency modes for upsert and delete APIs managing your vector collections.
You can apply attribute-based filtering to your upserted vectors. This allows you to restrict nearest-neighbor or BM25 query payloads to specific subsets of documents within a namespace.
No, the service uses a hosted embedding model, all-MiniLM-L6-v2, to generate 384-dimensional vectors. You simply embed your texts via the embed API and upsert the returned vectors into your namespace.
Use namespace management with server-generated UUIDs to create isolated vector collections. This keeps distinct document sets separate during upsert, compaction, and query operations.
discountedcookieStore 384-dimensional embeddings in Postgres and query cosine similarity with pgvector.
devMoezPerform vector similarity search with metadata filtering via the qdrant-client API.
zepdbIndex and search text with Zeppelin vector store and BM25 full-text indexing.
KarlinskySStore and search vector embeddings with metadata filtering via Python and JavaScript APIs.
