MindsDB Minor Release Notes v25.5.4.2

We’re excited to share the release of MindsDB v25.5.4.2 and also explore previous versions that feature major improvements to our documentation, enhanced SQL capabilities, deeper integration upgrades, and expanded functionality across our Knowledge Bases — all designed to make your experience smoother, smarter, and more powerful.

Documentation

We've significantly upgraded our documentation in v25.5.4.0 with improved navigation and a centralized knowledge hub. It now covers both Open Source and MindsDB Enterprise, offering a unified and user-friendly experience. In our MindsDB Open Source Documentation we have added new sections Connect, Unify and Respond to help you navigate more easily and find information faster. In v25.4.1 a new documents about the MCP are available, and the agent documentation has been improved for easier understanding.

Integrations

Our Writer integration now supports Agents bringing better data management, and MindsDB now extends to the cloud with the addition of SalesForce in v25.5.4.2!

To enhance security, we've implemented token-based authentication for the MCP server in v25.5.4.0. This new feature helps safeguard your data by providing more secure and controlled access, ensuring your information stays protected at all times

We’ve introduced a new Agent2Agent (A2A) server in v25.5.4.0 to enhance collaboration within the MindsDB ecosystem. This upgrade unlocks more interactive and dynamic capabilities, allowing AI agents to work together more seamlessly — making it easier to build, deploy, and scale intelligent workflows. We will officially launch this feature soon, so make sure to stay tuned! In v25.5.4.1, the A2A API was implemented with a better configuration and Docker setup for easier use. A2A configuration setup is now smoother thanks to improved logic behind it in v25.5.4.2.

Knowledge Base

Key improvements to the MindsDB knowledge base have been made to enhance usability and performance. We’ve added a new blueprint for connecting to LLM providers using a LiteLLM handler, making integration easier in v25.5.4.2. A default re-ranking model is now live, boosting the accuracy of search results. Plus, MindsDB models are no longer used for Knowledge Bases, simplifying the overall workflow. In v25.5.4.0 the updated documentation example makes it easier to understand how to work with the knowledge base, especially for new users. We’ve also added JSON chunking support, which optimizes how data is stored and retrieved — making interactions with large datasets faster and more efficient.

In a previous release, several key updates were implemented to improve the functionality and user experience of MindsDB’s knowledge base (KB). The documentation has been refined to provide clearer, more intuitive guidance. We’ve added support for AZURE_OPENAI_KEY, making it easier to embed models from Azure into your KBs. You can now run more complex queries, expanding the range of use cases you can tackle. To improve data structure, an ID column is now required, but to keep things simple, MindsDB will automatically generate one if it’s not provided — streamlining the setup process while ensuring consistency.

 SQL Operations

We’ve made several powerful enhancements to MindsDB’s SQL support to make your workflows more dynamic, flexible, and aligned with modern standards. You can now use variables directly in SQL, allowing for more adaptable query execution. We’ve also improved compatibility with industry-standard SQL features, including support for INSERT statements without column lists, offering greater flexibility in data operations. 

Additionally, inserts from SELECT queries are now registered in the queries table, making it easier to trace and optimize database performance. And with the new CREATE INDEX support within table-altering queries, you can now define indexing dimensions more efficiently — giving you better control over performance tuning.

Bug Fixes and Improvements

Our bug fixes and improvements enhance the overall reliability and flexibility of MindsDB. We’ve enhanced the release testing process, improved error messaging during staging, and refined file handling during uploads. Transformer usage is now optional, offering greater flexibility in AI workflows. Additionally, we’ve optimized data streaming from the data handler and resolved issues related to Handler Utilities for a smoother project startup and integration experience.

For a more detailed log of our release, you can check out our Release Notes v25.5.4.2 on Github.