@nodellmcache/chroma

Chroma vector-store adapter for NodeLLMCache. Implements VectorStoreAdapter, so it plugs into @nodellmcache/semantic-cache or works standalone.

Install

npm install @nodellmcache/chroma @nodellmcache/core chromadb
docker run -p 8000:8000 chromadb/chroma

Quick start

import { ChromaAdapter } from '@nodellmcache/chroma'

const store = new ChromaAdapter<{ source: string }>({
  path: 'http://localhost:8000',
  collection: 'docs', // created on first use (cosine)
})

await store.upsert('doc-1', embedding, { source: 'wiki' })
const matches = await store.query(queryEmbedding, 5, { source: 'wiki' })
// → [{ id: 'doc-1', score: 0.94, metadata: { source: 'wiki' } }]
await store.delete('doc-1')

Options

| Option | Default | Description | |--------|---------|-------------| | collection | — (required) | Collection name (created on demand) | | path | http://localhost:8000 | Server URL (or pass client) | | space | cosine | HNSW distance: cosine | l2 | ip | | maxRetries | 3 | Attempts per op on transient failure | | client | constructed | Inject an existing chromadb client (or compatible) |

Notes

• String ids are supported natively (no remapping).
• Scoring is 1 - distance; with the default cosine space that's cosine similarity (what SemanticCache expects).
• Metadata filters are flat equality; multiple keys are combined with Chroma's $and.

Testing

docker run -p 8000:8000 chromadb/chroma
CHROMA_URL=http://localhost:8000 pnpm --filter @nodellmcache/chroma test

License

MIT