ChromaDB Local Vector Store
Run Chroma as an embedded or client-server vector store. Generates collection setup, persistent storage configs, custom embedding functions, metadata filters, and migration paths to managed vector DBs later.
Chroma is the fastest way to prototype RAG locally without provisioning infra. This skill wires up persistent client storage, plugs in OpenAI/sentence-transformers embedding functions, handles collection upgrades, and writes the migration path to Pinecone or Weaviate when you outgrow it.
chromadb vector-database embeddings rag prototyping
When to use
Use during RAG prototyping, in local dev environments, for small-scale internal tools, or when you want zero-ops semantic search that ships with the app.
Examples
Persistent local RAG prototype
Embed and query docs entirely on disk
Set up Chroma with persistent storage and sentence-transformers embeddings to index our /docs folder for local RAG
Migrate from Chroma to Pinecone
Move beyond a local store when traffic grows
Write a migration script that exports our Chroma collection to Pinecone serverless, preserving metadata and IDs