Umami 3.4 Is Here: A Major Upgrade for Analytics, Automation, and Developers

Updated: September 2026

Umami 3.4.0 is now available, bringing a substantial set of new features and improvements to the open-source web analytics platform.

This release adds Annotations, Model Context Protocol (MCP) support, API key management, and a new typed API client, while also expanding analytics APIs, improving segments and visitor tracking, strengthening security, and fixing a wide range of database, UI, and mobile issues.

Whether you use Umami primarily to understand website traffic or build integrations around your analytics data, version 3.4 makes the platform considerably more capable.

Annotations: Put Your Analytics in Context

One of the most visible additions in Umami 3.4 is Annotations.

Analytics charts tell you what happened, but they don't always explain why. A traffic spike might coincide with a product launch, marketing campaign, newsletter, promotion, or other important event. Annotations let you record those events directly alongside your analytics.

You can now:

  • Add dated notes to website charts
  • Create, edit, and delete annotations
  • Create both all-day and time-specific notes
  • View chart markers for important events
  • Manage annotations with improved date-picker controls

This makes it easier to connect changes in your traffic and engagement with the events that may have influenced them.

Instead of maintaining a separate timeline of campaigns and launches, you can keep that context directly within your analytics.

Umami Gets Model Context Protocol Support

Umami 3.4 also introduces Model Context Protocol (MCP) support, opening up a new way to interact with analytics through compatible AI assistants.

MCP-compatible assistants can connect to Umami using read-only tools to explore analytics data. Available tools cover a broad range of Umami's reporting capabilities, including:

  • Traffic
  • Events
  • Sessions
  • Funnels
  • Goals
  • Journeys
  • Retention
  • Attribution
  • Revenue
  • Performance
  • Segments
  • Annotations

MCP can be accessed through a remote HTTP endpoint or locally through stdio using the Umami MCP package.

Authentication uses API keys and respects existing user and team permissions, providing a consistent access-control model.

For self-hosted installations, the MCP endpoint is disabled by default. Administrators can enable it by setting MCP_ENABLED=1, which makes the /mcp endpoint available.

The addition of MCP is particularly significant for teams building analytics workflows around AI assistants, since it provides a structured, read-only way for those tools to work with Umami data.

API Keys and a New Typed API Client

Version 3.4 also makes Umami's API more developer-friendly.

Users can now create and revoke account API keys from:

Settings → API keys

Alongside API key management, Umami introduces a new @umami/api-client TypeScript package generated directly from Umami's OpenAPI contracts.

The typed client provides:

  • Typed request parameters
  • Typed responses
  • Structured errors
  • ESM and CommonJS support
  • Generated OpenAPI documentation
  • Descriptions for API endpoints and fields

This gives developers a more structured way to build integrations while reducing the need to manually maintain API types.

Expanded Analytics APIs

The analytics API architecture has also received a significant update.

Umami 3.4 adds website-scoped GET endpoints for analytics queries and introduces dedicated resources and statistics endpoints for saved funnels and goals.

Performance data is now separated into dedicated datasets for:

  • Performance statistics
  • Performance charts
  • Performance metrics

There is also a dedicated UTM metrics endpoint.

For existing integrations, the release retains legacy report APIs through compatibility handlers. Existing saved report definitions and IDs are preserved as part of the API changes, helping reduce disruption for current users.

More Flexible Segments and Tracking

Segments receive an important improvement in 3.4.

Session property filters can now be saved as part of segments and combined with active filters. This provides more flexibility when creating reusable views of analytics data.

Visitor identification has also been expanded. The tracker now supports setting a visitor identity through the data-distinct-id attribute.

The release also adds support for two additional revenue currencies:

  • New Taiwan Dollar (TWD)
  • Icelandic Króna (ISK)

Umami 3.4 additionally introduces Georgian and Azerbaijani translations, expanding the platform's localization options.

Security Improvements

Security is another major focus of the release.

Umami 3.4 includes improvements to authentication, session handling, team access, and report permissions.

Among the changes are:

  • Rejection of partial 2FA tokens for authenticated API access
  • Binding authenticated sessions to password fingerprints
  • Exclusion of deleted teams from access checks
  • Invalidation of join codes for deleted teams
  • Prevention of website reassignment when updating reports
  • Gitleaks secret scanning
  • Updated security dependency overrides

These changes strengthen both the application's authentication model and its handling of access permissions.

Database and Analytics Fixes

The release also addresses several issues affecting analytics accuracy and database behavior.

Timezone inconsistencies between PostgreSQL and ClickHouse have been fixed, helping ensure more consistent analytics across supported database backends.

Session handling for identified visitors has also been improved. Visitors sharing the same IP address and user agent can now receive separate sessions based on their distinct ID.

For PostgreSQL, session activity queries now use correlated EXISTS queries, improving query behavior. Unnecessary sorting has also been removed from PostgreSQL visitor count queries.

These changes complement the new features by improving the underlying reliability and performance of analytics queries.

A Long List of UI and Usability Fixes

Umami 3.4 includes numerous smaller improvements throughout the interface.

The release fixes issues involving:

  • API client path parameters
  • Regular-expression filter values containing commas
  • Filter values when operators change
  • Custom page selector values
  • Wildcard controls in goal and funnel forms
  • Creating boards without descriptions
  • Boards referencing deleted websites
  • Pasting one-time passwords on mobile
  • Fullscreen date filters on mobile
  • Team navigation with configured base paths
  • The collapsed sidebar scrollbar
  • Full path metric label tooltips
  • Website selector behavior
  • Event detail wrapping
  • Chart tooltip styling

Individually, these may seem like small changes, but together they make the day-to-day analytics experience more consistent across desktop and mobile.

Updated Technology Stack

Umami 3.4 also updates several core dependencies and development tools.

The release includes:

  • Next.js 16.3.4
  • Prisma 7.10.0
  • React 19.3.0
  • React Zen component library upgrades
  • Shared theme tokens
  • Refined team and top navigation selectors
  • API integration coverage for PostgreSQL and ClickHouse
  • OpenAPI and generated-client validation in CI
  • Runtime, frontend, build tooling, and security dependency updates

These upgrades provide the foundation for the new functionality while keeping the project's underlying stack current.

Automatic Migrations

Umami 3.4 introduces database migrations for the new annotations and API key functionality.

The release includes:

  • prisma/migrations/25_add_annotation
  • prisma/migrations/26_add_api_key

Migrations run automatically during the build process.

Existing saved report definitions and IDs are retained despite the analytics API changes, which helps preserve existing reporting configurations.

What Umami 3.4 Means for Users

Umami 3.4 is more than a collection of incremental fixes. It expands the ways people can use Umami and the ways developers can build around it.

Annotations make analytics easier to interpret by adding important events directly to charts. MCP support introduces a new interface for AI-assisted, read-only analytics exploration. API keys and the typed API client make integrations easier to build and maintain, while the expanded analytics endpoints provide more focused access to specific datasets.

At the same time, improvements to segments, visitor identification, security, database queries, localization, and the user interface make the platform more robust across a wide range of use cases.

For existing Umami installations, version 3.4 represents a broad upgrade across the product. For developers, it provides new building blocks for integrations and automation. And for analytics users, it adds more context, flexibility, and ways to work with their data.

Umami 3.4 brings analytics, developer tooling, and AI interoperability closer together — while continuing to refine the core analytics experience that Umami is built around.

Get Started with Managed Umami

Ready to try Umami without the DevOps hassle? UmamiEngine provides fully managed Umami analytics that runs the latest Umami version, including the new features described above such as annotations, MCP support, and the expanded analytics APIs. No servers, no databases, no maintenance, just privacy-first analytics in minutes. Get started today and take control of your analytics data.

Get fully managed Umami in minutes