@radzor/vector-search

Vector embedding storage and semantic search. Supports Pinecone, Qdrant, and pgvector. Index documents with embeddings and query by semantic similarity for RAG pipelines, recommendation engines, and duplicate detection.

AI & MLv0.1.0typescriptServervectorembeddingssemantic-searchragpineconeqdrantpgvectorsimilarityby Radzor

Install

View source on GitHub →

$npx radzor@latest add vector-search

⚠ Constraints: Vectors must be pre-computed — this component does not generate embeddings. Use @radzor/llm-completion or the OpenAI embeddings API to generate vectors before indexing. Dimensions must be set at index creation time and cannot be changed.

Inputs

Name	Type	Default	Description
provider*	'pinecone' \| 'qdrant' \| 'pgvector'	—	Vector database provider.
apiKey	string	—	API key for Pinecone. Not required for Qdrant or pgvector.PINECONE_API_KEY
host	string	—	Host URL for Qdrant or pgvector connection string. Not used for Pinecone.
indexName*	string	—	Name of the index / collection / table to use.
dimensions*	number	—	Embedding vector dimensions. Must match the embedding model (e.g. 1536 for text-embedding-3-small).
metric	'cosine' \| 'euclidean' \| 'dotProduct'

{ "llm": { "constraints": "Vectors must be pre-computed — this component does not generate embeddings. Use @radzor/llm-completion or the OpenAI embeddings API to generate vectors before indexing. Dimensions must be set at index creation time and cannot be changed.", "usageExamples": "llm/examples.md", "integrationPrompt": "llm/integration.md" }, "name": "@radzor/vector-search", "tags": [ "vector", "embeddings", "semantic-search", "rag", "pinecone", "qdrant", "pgvector", "similarity" ], "events": [ { "name": "onIndexed", "payload": { "id": "string", "dimensions": "number" }, "description": "Fired when a vector is successfully indexed." }, { "name": "onSearchComplete", "payload": { "query": "string", "durationMs": "number", "resultCount": "number" }, "description": "Fired when a search query completes." }, { "name": "onError", "payload": { "code": "string", "message": "string", "provider": "string" }, "description": "Fired when an operation fails." } ], "inputs": [ { "name": "provider", "type": "'pinecone' | 'qdrant' | 'pgvector'", "required": true, "description": "Vector database provider." }, { "name": "apiKey", "type": "string", "envVar": "PINECONE_API_KEY", "required": false, "description": "API key for Pinecone. Not required for Qdrant or pgvector." }, { "name": "host", "type": "string", "required": false, "description": "Host URL for Qdrant or pgvector connection string. Not used for Pinecone." }, { "name": "indexName", "type": "string", "required": true, "description": "Name of the index / collection / table to use." }, { "name": "dimensions", "type": "number", "required": true, "description": "Embedding vector dimensions. Must match the embedding model (e.g. 1536 for text-embedding-3-small)." }, { "name": "metric", "type": "'cosine' | 'euclidean' | 'dotProduct'", "default": "cosine", "required": false, "description": "Distance metric for similarity search." }, { "name": "topK", "type": "number", "default": 10, "required": false, "description": "Default number of results to return." } ], "radzor": "1.0.0", "actions": [ { "name": "index", "params": [ { "name": "id", "type": "string", "description": "Unique identifier for the vector." }, { "name": "vector", "type": "number[]", "description": "Embedding vector." }, { "name": "metadata", "type": "Record<string, unknown>", "required": false, "description": "Arbitrary metadata stored alongside the vector." } ], "returns": "Promise<void>", "description": "Store a vector with its metadata." }, { "name": "indexBatch", "params": [ { "name": "items", "type": "Array<{ id: string; vector: number[]; metadata?: Record<string, unknown> }>", "description": "Batch of vectors to index." } ], "returns": "Promise<void>", "description": "Index multiple vectors in a single request." }, { "name": "search", "params": [ { "name": "vector", "type": "number[]", "description": "Query embedding vector." }, { "name": "topK", "type": "number", "required": false, "description": "Number of results to return." }, { "name": "filter", "type": "Record<string, unknown>", "required": false, "description": "Metadata filter (provider-specific syntax)." } ], "returns": "Promise<SearchResult[]>", "description": "Find the most semantically similar vectors." }, { "name": "delete", "params": [ { "name": "id", "type": "string", "description": "ID of the vector to remove." } ], "returns": "Promise<void>", "description": "Delete a vector by ID." }, { "name": "getStats", "params": [], "returns": "Promise<IndexStats>", "description": "Get index statistics." } ], "outputs": [ { "name": "searchResults", "type": "SearchResult[]", "fields": [ { "name": "id", "type": "string" }, { "name": "score", "type": "number" }, { "name": "metadata", "type": "Record<string, unknown>" } ], "description": "Array of matched documents with id, score, and metadata." }, { "name": "indexStats", "type": "IndexStats", "fields": [ { "name": "totalVectors", "type": "number" }, { "name": "dimensions", "type": "number" }, { "name": "fullness", "type": "number", "required": false }, { "name": "indexName", "type": "string" } ], "description": "Index statistics: total vectors, dimensions, fullness." } ], "runtime": "server", "version": "0.1.0", "category": "ai", "languages": [ "typescript" ], "description": "Vector embedding storage and semantic search. Supports Pinecone, Qdrant, and pgvector. Index documents with embeddings and query by semantic similarity for RAG pipelines, recommendation engines, and duplicate detection.", "dependencies": { "radzor": [ "@radzor/llm-completion" ], "packages": { "@pinecone-database/pinecone": "^3.0.0" } }, "composability": { "connectsTo": [ { "output": "searchResults", "compatibleWith": [ "@radzor/llm-completion.action.complete.prompt" ] } ] } }

@radzor/vector-search

Install

Inputs

Outputs

Actions

Events

Composability

radzor.manifest.json

Version History

Name	Type	Description
searchResults	SearchResult[]	Array of matched documents with id, score, and metadata. .idstring .scorenumber .metadataRecord<string, unknown>
indexStats	IndexStats	Index statistics: total vectors, dimensions, fullness. .totalVectorsnumber .dimensionsnumber .fullnessnumber(optional) .indexNamestring