Kraken
This release brings enhanced AI model configuration controls, improved AI··Room experience, fullscreen editing updates, and fixes for critical deployment and task configuration issues. The kraken was originally described by sailors as being so massive that it could be mistaken for an island, and when it submerged, it would create dangerous whirlpools that could sink entire ships
CHANGES
Section titled “CHANGES”Improved AI··Room experience 🏠
Section titled “Improved AI··Room experience 🏠”Ref: #10839, #10853, #10968
Experience a smoother and more refined AI··Room with our comprehensive upgrades:
-
Multiple services can now be called in a single exchange
.DzAyXN5t_ZMCVkm.png)
-
Better visual alignment with properly positioned icons
-
Better JSON support
.BKMPLEZf_Z2jKHGp.png)
.BlEkERW8_Z1hQJl5.png)
-
Cleaner interface with unnecessary elements removed
-
Faster and more responsive real-time updates thanks to optimized audit event processing
-
Enhanced markdown rendering system for more reliable message display stability when switching between chatrooms
Enhanced AI··Task settings ⚙️🤖
Section titled “Enhanced AI··Task settings ⚙️🤖”Ref: #10485, #11015
We now provide direct access to low-level AI model configuration parameters. A dedicated settings tab displays the specific parameters available for your selected model, allowing technical users to fine-tune advanced settings when needed.
The available configuration options are dependent on the specific LLM you choose, ensuring you only see settings that apply to your model.
These improvements provide complete access to all configuration parameters, giving you full control to fine-tune how the LLM is configured at a technical level.
Here are some examples:


Enhanced fullscreen editing with accessible menu 🚀
Section titled “Enhanced fullscreen editing with accessible menu 🚀”Ref: #10651
The menu is now available in fullscreen mode - something we had forgotten to include previously.

Improved Auditability 📋
Section titled “Improved Auditability 📋”Ref: 10854
The audit‑logs API now provides more detailed information. New audit events include credential/get_completed: Successfully retrieved the credential value
API documentation is available at: https://docs.mindflow.io/guides/administrating-the-platform/exporting-logs
- Fixed tooltip displaying PARAM-* instead of field name for required fields in error messages (#10961)
- Dry Run does not take into account recent step modifications (#10983)
- AI··Task Model ID Disappears After Page Refresh (#10981)
- Cannot create a working Flowhook without parameters. (#10994)
- Model ID is undefined when creating AI··Task (#11004)
- SMTP does not work for customer-owned domains when running flows (#11011)
- Users cannot dry run a step in a versioned flow deployment if the step doesn’t exist in the latest version (#11048)
- Loss of focus during typing in additional properties (#11093)
- Fixed infinite loop in ForEach step that occurred without generating error logs (#11109)
- FX is unavailable within loop step conditions (#11137)