Version 6.6.5952 dated 09.06.2026

⚠️ IMPORTANT BEFORE UPDATING — PostgreSQL with the pgvector extension is required

Starting with this version, the application requires a PostgreSQL database with the pgvector extension installed — it is used by the AI chat knowledge base and vector search.

Without the extension, the application will not start. On startup, the database migration performs CREATE EXTENSION vector; if the extension is missing, the container terminates with an error — there is no "graceful degradation" mode.

What to do before updating: install the pgvector extension in your PostgreSQL. There are several ways to do this — commands and important nuances are collected in a separate guide:

➡️ Installing pgvector in PostgreSQL

This requirement appeared along with the AI chat knowledge base. Versions up to and including 6.6.5694 did not require the extension — when upgrading from them, changing the PostgreSQL image/configuration is mandatory.

Release information

Parameter	Value
Version	6.6.5952
Date	09.06.2026
Commit	44a94ee5c

✨ New Features

• The process modeling interface now includes a simulation results panel with a clear overview of key KPIs, process "health," and a list of issues — making it easier to understand where the process performs well and where time and resources are lost.
• A 2×2 KPI grid and a process "health" indicator (an arc with sub-scores) have been added to the simulation results, helping to quickly identify bottlenecks and prioritize improvements.
• The diagram editor now offers a new data view in "chip-pivot" format, with the option to switch to a flat list — simplifying the analysis of complex models and related objects.
• The built-in meeting booking widget now operates on Storm-CRM instead of a third-party service: customers can book directly from the website, and all requests are immediately sent to your CRM.
• The modeling interface has improved tools for precise control of element geometry and automatic diagram layout, allowing you to quickly arrange diagrams into a neat and readable format.
• The admin panel now allows restoring roles and org structure elements if they were deleted or corrupted, reducing the risk of losing important access settings.

🚀 Improvements

• It is now easier to roll back to previous versions of diagrams in the AI chat and diagram editing module: this increases the safety of experiments with process models and speeds up returning to proven variants.
• Marketing pages and forms (including hints for embedding forms/iframe) have become more flexible in design and better fit the company's brand style.
• Automatic layout of BPMN diagrams has become neater and more predictable, reducing manual refinement of diagrams after automatic placement of elements.
• Simulation processes can be started, their status tracked in real time, and quickly restarted based on a previous run, which accelerates the "modeled — viewed the result — improved" cycle.

🔧 Fixes

• Resolved issues where reloading the same data via the public API could result in duplicate org structure elements.
• Improved subscription filtering and extension logic: manual invoices and the transition from the trial period to a paid plan are now handled more accurately, avoiding unexpected scenarios for clients.
• Enhanced AI chat stability: reduced the number of false refusals in responses and improved search within the knowledge base so that relevant materials are found by keywords.
• Fixed rare cases of "silent" link loss during batch operations with roles and org structure elements, improving the reliability of administration.
• Increased the resilience of the integration with the external AI connector: restarting client applications no longer causes failures in scenario execution.

Disclaimer

This changelog is generated automatically using AI based on developer commits. Some of the described features may not be available in your configuration or may require additional licenses. For details, please refer to the documentation or contact support.

Docker image

docker pull cr.selcloud.ru/stormbpmn-enterprise/stormbpmn_fullstack_enterprise:6.6.5952

Image signature verification (Cosign)

All Docker images are cryptographically signed using Sigstore Cosign. To verify the authenticity of an image:

1. Install Cosign:

# macOS
brew install cosign

# Linux
curl -sSL https://github.com/sigstore/cosign/releases/latest/download/cosign-linux-amd64 -o /usr/local/bin/cosign
chmod +x /usr/local/bin/cosign

2. Verify the signature:

cosign verify \
  --certificate-identity-regexp="https://github.com/KotskinKotskin/stormfullstack/*" \
  --certificate-oidc-issuer="https://token.actions.githubusercontent.com" \
  cr.selcloud.ru/stormbpmn-enterprise/stormbpmn_fullstack_enterprise:6.6.5952

A successful verification guarantees that the image was built in the official CI/CD pipeline and has not been tampered with.