Attu - The AI Workbench for Milvus

Attu is an AI-native management tool for Milvus vector databases. Connect to multiple Milvus clusters from a single instance, browse collections, run vector searches, manage backups, monitor health, and chat with an AI agent that understands your data.

Available as a web app (Docker / Kubernetes) or desktop app (macOS, Linux, Windows).

---

What's New in v3

Attu v3 is a ground-up rewrite with a modern stack (TanStack Start, React 19, Tailwind CSS v4, shadcn/ui) and major new capabilities:

• AI Agent - Chat with an LLM-powered assistant that can query collections, manage schemas, and execute Milvus operations through natural language. Supports custom skills and tool-call approval workflows.
• Backup & Restore - Create full or incremental backups, download as ZIP, upload to restore. Supports S3, MinIO, GCS, and Azure Blob Storage.
• Tasks & Data Movement - Track long-running import, export, backup, restore, and copy tasks with retry/cancel controls.
• Embedding Configuration - Configure embedding providers directly in Attu for vector search with text input.
• REST API Playground - Interactive API testing with database/collection context pickers.
• Redesigned Explorer - Hierarchical database/collection browser with schema builder, data import/export (CSV, JSON, Parquet), vector search, snapshots, external collections, partition management, and segment viewer.
• Cluster Administration - Topology visualization, role-based access control with cross-cluster role copying, scoped collection access, user management, configuration viewer, metrics, slow request analysis, diagnostics, environments, and resource groups.
• Dark/Light Theme - Geist font, Attu green primary palette, designed for long working sessions.

---

Quick Start

Docker (recommended)

docker run -d --name attu \
  -p 3000:3000 \
  -e MILVUS_ADDRESS=host.docker.internal:19530 \
  -v attu-data:/data \
  zilliz/attu:v3.0.0-beta.6

Open http://localhost:3000 and connect to Zilliz Cloud or your open-source Milvus 3.x instance.

The Docker image stores its SQLite database at /data/attu.db by default. The -v attu-data:/data volume persists your saved connections, agent conversations, and preferences across container restarts.

Docker Compose (Milvus + Attu)

services:
  milvus:
    # Attu v3.0 supports open-source Milvus 3.x. Do not use Milvus 2.6.x here.
    image: milvusdb/milvus:<supported-3.x-tag>
    ports:
      - "19530:19530"
      - "9091:9091"
    command: milvus run standalone
    volumes:
      - milvus-data:/var/lib/milvus

  attu:
    image: zilliz/attu:v3.0.0-beta.6
    ports:
      - "3000:3000"
    environment:
      - MILVUS_ADDRESS=milvus:19530
    volumes:
      - attu-data:/data
    depends_on:
      - milvus

volumes:
  milvus-data:
  attu-data:

docker compose up -d

The Docker Compose example is for open-source Milvus 3.x compatibility validation. Attu v3.0 also supports Zilliz Cloud, but it does not support open-source Milvus 2.6.x or earlier Milvus 2.x releases.

Desktop App

Download the latest release for your platform:

Platform	Download
macOS (Apple Silicon)	.dmg
Linux	.AppImage / .deb
Windows	.exe

macOS note: If you see "attu.app is damaged and cannot be opened", run:

sudo xattr -rd com.apple.quarantine /Applications/Attu.app

---

Features

Multi-Cluster Management

Connect to multiple Milvus instances from a single Attu deployment. Add, edit, and switch between connections in the sidebar. Each cluster has its own dedicated workspace with independent monitoring, agent sessions, and preferen