Milvus DB Studio

A friendly GUI for the Milvus vector database — one command away

Browse collections · Insert & view records · Run vector search · Query metadata · Send raw REST calls.

What is this?

Milvus DB Studio is a visual dashboard for Milvus — an open-source database built for storing and searching AI vectors (embeddings).

Instead of typing long curl commands to inspect your data, you get a clean web UI that opens right in your browser. Think of it like a "phpMyAdmin" or "pgAdmin", but for Milvus.

You don't install anything permanently. One command downloads and runs it:

npx milvus-db-studio

New to npx? It's a tool that ships with Node.js. It downloads a package, runs it once, and cleans up after itself — so there's nothing to install or uninstall.

What you can do with it

• Browse collections — see all your collections, their schema, and load status
• Manage records — insert, view, and delete data without writing code
• Vector search — find the most similar vectors (the core of AI/RAG apps)
• Scalar query — filter records by metadata fields
• Raw REST console — send any Milvus API request directly
• Local or Cloud — connect to a Milvus server on your machine or Zilliz Cloud

Before you start (Prerequisites)

You only need two things:

1. Node.js 18 or newer — download here. Check with:

   node --version

2. A running Milvus server — don't have one? No problem, the next section shows you how to start one for free in under a minute.

Step 1 — Start a Milvus server (local)

If you already have a Milvus server running, skip to Step 2.

The easiest way to run Milvus locally is with Docker. If you don't have Docker yet, install Docker Desktop first.

# 1. Download the official Milvus setup file
wget https://github.com/milvus-io/milvus/releases/download/v3.0-beta/milvus-standalone-docker-compose.yml -O docker-compose.yml

# 2. Start Milvus in the background
docker compose up -d

That's it! Milvus is now running and listening at:

http://localhost:19530

Tip: To stop Milvus later, run docker compose down in the same folder.

Step 2 — Open Milvus DB Studio

In your terminal, run:

npx milvus-db-studio

Your browser opens automatically with the Studio UI. 🎉

Step 3 — Connect to your server

In the Studio UI, enter these settings:

| Setting | Value | | --- | --- | | Connection type | Local | | Endpoint URI | http://localhost:19530 |

Click Connect and you're in.

Step 4 — Create your first collection

A collection is like a table in a normal database. Let's create one called demo_collection that stores a small 3-dimension vector plus some text.

You can do this two ways:

Option A — Use the Studio UI (recommended for beginners): open the Resources tab and fill in the create-collection form.

Option B — Use the terminal (copy & paste this):

curl -X POST http://localhost:19530/v2/vectordb/collections/create \
  -H "Content-Type: application/json" \
  -d '{
    "collectionName": "demo_collection",
    "schema": {
      "autoId": false,
      "enabledDynamicField": true,
      "fields": [
        { "fieldName": "id", "dataType": "Int64", "isPrimary": true },
        { "fieldName": "vector", "dataType": "FloatVector", "elementTypeParams": { "dim": "3" } },
        { "fieldName": "text", "dataType": "VarChar", "elementTypeParams": { "max_length": 512 } }
      ]
    },
    "indexParams": [
      { "fieldName": "vector", "metricType": "COSINE", "indexName": "vector", "indexType": "AUTOINDEX" }
    ]
  }'

What this means: id is the unique key, vector holds the AI embedding (3 numbers here for simplicity), and text stores a label. COSINE is how similarity is measured.

Step 5 — Add some data

Now insert a few example rows:

curl -X POST http://localhost:19530/v2/vectordb/entities/insert \
  -H "Content-Type: application/json" \
  -d '{
    "collectionName": "demo_collection",
    "data": [
      { "id": 1, "vector": [0.1, 0.2, 0.3], "text": "hello milvus" },
      { "id": 2, "vector": [0.2, 0.3, 0.4], "text": "vector search" },
      { "id": 3, "vector": [0.9, 0.8, 0.7], "text": "ai native database" }
    ]
  }'

Switch back to the Studio UI, refresh, and open demo_collection — your records are there, ready to browse, query, and search. ✅

Connecting to Zilliz Cloud (instead of local)

Using a hosted Zilliz Cloud cluster? Pass your cluster URL and API key:

npx milvus-db-studio --url https://my-cluster.zillizcloud.com --api-key YOUR_API_KEY

This automatically switches the UI into Cloud mode and pre-fills your details.

Command-line options (advanced)

You can pre-fill connection settings and tweak how the server runs. All flags are optional.

| Option | What it does | Default | | --- | --- | --- | | -u, --url <url> | Milvus endpoint shown in the UI | http://localhost:19530 | | --database <name> | Pre-fill the database name | — | | --token <token> | Pre-fill a token (local auth, e.g. user:password) | — | | --api-key <token> | Pre-fill an API key (switches to Cloud mode) | — | | --username <name> | Pre-fill username | — | | --password <pass> | Pre-fill password | — | | --api-key-header <h> | Custom auth header name | Authorization | | -p, --port <port> | Port to serve the UI on | first free from 7070 | | --host <host> | Host to bind to | 127.0.0.1 | | --no-open | Don't open the browser automatically | — | | -h, --help | Show all options | — | | -v, --version | Show the version number | — |

Examples:

# Just launch it
npx milvus-db-studio

# Use a custom endpoint and port
npx milvus-db-studio --url http://localhost:19530 --port 9000

# See every option
npx milvus-db-studio --help

Troubleshooting

The browser didn't open / I can't connect. Make sure your Milvus server is running (docker compose up -d) and that the Endpoint URI in the UI matches http://localhost:19530.

Port already in use. Pick another port: npx milvus-db-studio --port 9000.

npx says command not found. Install Node.js 18+ from nodejs.org, then try again.

Is my data safe? Everything runs locally on your machine. Connection details stay in your browser; nothing is sent anywhere except the Milvus server you point at.

License

MIT © Harish Kaparwan