mongoui

Getting started

Getting started

Install mongoui, connect to MongoDB, choose optional AI, and run your first database workflow.

mongoui is a local-first desktop app for macOS and Windows. It runs on your machine and talks directly to MongoDB. Full documents, connection strings, and credentials are never sent to mongoui.

Install

  1. Download the build for your platform from the download section. On macOS, pick Apple Silicon or Intel (the page recommends one based on your Mac). On Windows, grab the 64-bit installer.
  2. macOS: open the .dmg and drag mongoui to Applications. The build is signed and notarized by Apple, so it opens without a Gatekeeper warning.
  3. Windows: run the installer. The current Windows build is unsigned, so SmartScreen may prompt. Choose More info, then Run anyway.

Connect

On first launch you'll land on the Connect screen. Paste a standard MongoDB connection string:

mongodb://127.0.0.1:27017

For a local replica set (the recommended local setup), add ?directConnection=true. See Connecting to MongoDB for the details on URIs, Atlas, and authentication.

Choose Read only when you want an application-level safeguard against database changes. The connection screen still lets you test MongoDB before opening the workspace.

Add AI only when you need it

The database connection is the only requirement. AI is optional. Without a provider, you can browse and edit documents, run queries and pipelines, inspect performance, compare environments, monitor change streams, and use every manual maintenance workflow.

To generate queries or use the AI Database Operator, select Anthropic, OpenAI, Google Gemini, Ollama, or a custom OpenAI-compatible endpoint. Hosted providers need your API key. Ollama runs on loopback and needs no key. Choose Test AI before you connect, or configure the provider later in Settings. See Configure an AI provider.

Run your first Smart Scan

Once connected, open Smart Scan and run it. In one pass mongoui checks:

  • Indexes: which ones earn their keep, which only cost writes
  • Storage: where space goes, and what's reclaimable
  • Duplicates: clusters of documents that match on a key you choose
  • Schema: field shapes, type drift, and missing fields

Anything that needs attention shows up with a specific, plain-English explanation and a safe path to fix it. Nothing is ever written without a dry run first.

Pick the workflow that matches the job

  • Use Browse to run a find, draft a query from plain English, edit one document, or dry-run a bulk update.
  • Use Optimize to analyze an explain() plan or draft a read-only aggregation.
  • Use Operator to turn a broader database task into a typed plan with separate read, preview, and write approvals.
  • Use Profiler history to inspect an existing system.profile collection without changing profiler configuration.
  • Use Environment Compare to compare one collection or a whole deployment through a separate read-only destination connection.

Start with natural-language queries, the AI Database Operator, or the full list on the documentation index.