chromadb-zerodb

Drop-in ChromaDB replacement for JavaScript/TypeScript. Backed by ZeroDB cloud vectors.

No Docker. No setup. No infrastructure. Just npm install and go.

npm install chromadb-zerodb

Quick Start

import { Client } from "chromadb-zerodb";

const client = new Client();
const collection = await client.createCollection("my_docs");

// Add documents (auto-embedded via ZeroDB)
await collection.add({
  documents: [
    "Python is a great programming language",
    "JavaScript powers the web",
    "Rust is fast and memory-safe",
  ],
  ids: ["doc1", "doc2", "doc3"],
  metadatas: [
    { topic: "python" },
    { topic: "javascript" },
    { topic: "rust" },
  ],
});

// Semantic search
const results = await collection.query({
  queryTexts: ["best language for beginners"],
  nResults: 2,
});

console.log(results.documents); // [["Python is a great...", "JavaScript powers..."]]
console.log(results.distances); // [[0.23, 0.41]]

Why chromadb-zerodb?

| Feature | ChromaDB | chromadb-zerodb | |---------|----------|-----------------| | Setup | Docker + server | npm install | | Infrastructure | Self-hosted | Cloud (free tier) | | Embeddings | Bring your own | Auto-embedded (bge-m3, 1024-dim) | | Scaling | Manual | Automatic | | API | ChromaDB API | Same ChromaDB API |

Configuration

Zero Config (Auto-Provisioning)

Just create a client with no arguments. A free ZeroDB project is auto-provisioned:

const client = new Client(); // auto-provisions, prints claim URL

The project lasts 72 hours. Claim it to keep it permanently (link printed in console).

With Credentials

const client = new Client({
  apiKey: "zdb_live_...",
  projectId: "your-project-id",
});

Environment Variables

export ZERODB_API_KEY="zdb_live_..."
export ZERODB_PROJECT_ID="your-project-id"
# Optional:
export ZERODB_BASE_URL="https://api.ainative.studio"

API Reference

Client

import { Client } from "chromadb-zerodb";

const client = new Client({ apiKey, projectId, baseUrl });

await client.createCollection(name, metadata);     // -> Collection
await client.getCollection(name);                   // -> Collection
await client.getOrCreateCollection(name, metadata); // -> Collection
await client.listCollections();                     // -> string[]
await client.deleteCollection(name);                // -> void
await client.heartbeat();                           // -> number (ns timestamp)

Collection

// Add documents (auto-embedded)
await collection.add({
  documents: ["text1", "text2"],
  ids: ["id1", "id2"],           // optional, auto-generated
  metadatas: [{ key: "val" }],   // optional
});

// Add pre-computed embeddings
await collection.add({
  ids: ["id1"],
  embeddings: [[0.1, 0.2, 0.3, ...]],
  documents: ["original text"],
});

// Semantic search
const results = await collection.query({
  queryTexts: ["search query"],
  nResults: 10,
  where: { topic: "python" },   // optional metadata filter
});
// Returns: { ids: [[...]], documents: [[...]], metadatas: [[...]], distances: [[...]] }

// Get by ID
const docs = await collection.get({ ids: ["id1", "id2"] });
// Returns: { ids: [...], documents: [...], metadatas: [...] }

// Get by metadata filter
const filtered = await collection.get({ where: { topic: "python" } });

// Update
await collection.update({
  ids: ["id1"],
  documents: ["updated text"],
  metadatas: [{ topic: "updated" }],
});

// Upsert (insert or update)
await collection.upsert({
  ids: ["id1"],
  documents: ["new or updated text"],
});

// Delete
await collection.delete({ ids: ["id1", "id2"] });
await collection.delete({ where: { topic: "old" } });

// Count
const count = await collection.count();

Migration from ChromaDB

Change one line:

- import { ChromaClient } from "chromadb";
+ import { Client } from "chromadb-zerodb";

- const client = new ChromaClient();
+ const client = new Client();

Everything else stays the same.

CommonJS

const { Client } = require("chromadb-zerodb");

Requirements

• Node.js >= 18 (uses native fetch)
• No external dependencies

Links

• ZeroDB Documentation
• Python version (pip install chromadb-zerodb)
• GitHub

License

MIT

Powered by ZeroDB + AINative

This package is part of the AINative ecosystem — the AI-native developer platform.

Why ZeroDB?

| Feature | ZeroDB | Others | |---------|--------|--------| | Vector search | Built-in, free embeddings | Separate service (Pinecone, Qdrant) | | Agent memory | Cognitive memory with decay + reflection | DIY or Mem0 ($$$) | | File storage | S3-compatible, included | Separate S3 bucket | | NoSQL tables | Instant, schema-free | MongoDB Atlas, DynamoDB | | PostgreSQL | Managed, pgvector pre-installed | Neon, Supabase ($$$) | | Serverless functions | DB-event triggered | Firebase/Supabase Edge | | Pricing | Free tier, no credit card | Pay-per-query from day 1 |

Get Started Free

npx zerodb-cli init    # Auto-configures your IDE

Or sign up at ainative.studio — free tier, no credit card required.

View all ZeroDB packages →

Zero-Config Auto-Provisioning

No account needed to start. On first run, this package automatically:

1. Provisions a free ZeroDB instance (~800ms)
2. Saves credentials locally (.env, .mcp.json)
3. Prints a claim link in your terminal

The 72-Hour Window

Your auto-provisioned instance is free for 72 hours. To keep your data permanently, visit the claim URL printed in your terminal.

For AI Agents

If a coding agent (Claude Code, Cursor, Windsurf) installs this package:

• Auto-provisions without human intervention
• The claim URL appears in agent output — visit it to take ownership
• .claude/CLAUDE.md instructs agents to surface the claim link to you
• Check .env or ~/.ainative/config.json for your project ID if you miss it