August was a month of giving teams more control over their own workspace, more room to trust what they automate, and a quieter but real security and reliability pass underneath it.
Test cases get a real version history instead of silent overwrites. Project admins get to clean up their own taxonomy instead of filing a ticket. Playwright finally reaches the private networks the rest of the platform already could, and a critical vulnerability in the Katalon Agent gets patched along the way.
Katalon AI Assistant asks before it touches anything in your connected tools, and can now read your own project documents when you want it to. TrueTest opens up to MCP clients, and Katalon Studio 11.5.0 shipped alongside all of it. Here's everything that shipped.
August's TestOps and TestCloud work is about giving teams direct control over things that used to require a workaround or a support request, plus a heavier-than-usual reliability and security pass underneath.
Every manual test case now keeps a version history. Each saved change creates a new, immutable version, nothing is overwritten, and every test run permanently records the exact version it executed, so past results never change when the test case is edited later. Saving a change creates the next draft minor and switches the test case back to Draft status: editing v2 creates v2.1 (Draft), and publishing promotes it to the next major version, so publishing after v2.3 creates v3 (Published).
Only the latest version is editable; older versions are read-only and can be restored as a new latest version. Any version can be executed, including draft minors, using the latest, latest published, or a specific-version option at run creation, and the run is stamped with whichever version it used.
From Project Configuration, authorized teams can review usage counts and manage Feature Areas, Test Types, and Tags directly: add and rename values, and consolidate typo or case-variant duplicates, "Regression," "regression," and "Regresion" becoming one value, through a merge workflow, without a data migration request or permanent duplicate taxonomies scattered across test cases.
Deleting a value no longer requires manually tracking down and fixing your own saved filters that referenced it; that cleanup now happens automatically.
The Tags table itself is also cleaner: since tags in Gen 3 projects can only be assigned to test cases, it no longer shows Test Runs and Test Results usage columns that never applied.
And if you use custom roles, a role with Read permission for Project System Configuration now gets access as configured, a bug that previously denied it under the newer navigation experience is fixed.
A platform-triggered Playwright run honors your existing Katalon Tunnel and IP Whitelist selection. Nothing changes in how a tunnel or allowlist is configured. A large share of the enterprise base tests internal applications, a UAT behind a corporate network, staging on a private subnet, production behind a firewall, and until now those teams had to run Playwright themselves to reach them.
A handful of supporting fixes ship alongside it: API test cases now pass correctly when Private/Local Testing is set to Tunnel and "Also include API Calls" is checked, a Playwright repository refresh now removes suites that were deleted upstream and creates newly added execution profiles, editing a Playwright schedule's Advanced Settings now shows only the options that actually apply to Playwright, and Playwright tunnel runs on Linux now succeed as reliably as they do on Windows.
Linux browser executions can now create network-log artifacts when katalon:options.enableNetwork=true, using the same Network view and download flow already available for Windows and macOS, and video recording for Linux executions is fixed. TestCloud also supports Linux browser versions 146 through 151 as execution environments, with Katalon Framework compatibility covered. Diagnosing a Linux failure no longer means re-running it on a different platform just to get the evidence, and teams stay current with browser and OS releases.
How to view network and performance data • View TestOps release notes
Katalon Studio also shipped a major release this month: redesigned PDF reports with AI-generated failure analysis, Katalon AI Assistant chat memory that persists across sessions, built-in AI keywords for prompts and assertions, Android App Bundle support, project-wide search and Microsoft Entra ID authentication for MCP clients, and a performance pass that resolves the large-project slowdowns some teams hit in 11.3.0 and 11.4.0.
What's New in Katalon Studio 11.5.0 • View 11.5.0 release notes
Upload your team's documents, requirements, specs, test plans, style guides, domain glossaries, into a project knowledge base, so the material that defines how your product should behave lives with the project it belongs to.
Document Storage accepts PDF, DOCX, XLSX, PPTX, TXT, CSV, MD, PNG, JPG, and JPEG files up to 100 MB, shows each upload's processing status through to Completed, prevents duplicate filenames, and lets you search and filter what the project holds.
Uploading, processing, and managing documents are available now; retrieval of that content inside Katalon AI Assistant answers and generated tests is rolling out separately and is not yet enabled in every project, so verify it in your own project before relying on it.
Every write to a connected tool is now something you approve. Before Katalon AI Assistant writes to a Jira issue or an Azure DevOps work item, it raises an approval request naming the tool, the action, the target, and the exact content it intends to send, and it acts only after you confirm.
Nothing is auto-approved until you choose to always allow a specific tool, actions that could delete or irreversibly overwrite your data are switched off unless explicitly enabled, and declining ends the request cleanly instead of surfacing an error.
When a manual test case calls another test case, the AI test runner resolves that reference and runs the called steps as part of the execution instead of skipping over them, so teams that factor shared login, setup, or checkout steps into reusable test cases get a result that reflects the whole flow.
Clarification questions and generated-test review render directly in the product instead of relying on embedded MCP App views, so you can answer questions in context and review, save, or discard generated test cases through the native experience, with fewer handoffs and clearer recovery from invalid generation inputs.
Eligible enterprise teams using an organization AI key can reach a private Azure OpenAI or OpenAI-compatible endpoint through an existing Katalon Tunnel, without exposing that endpoint publicly or changing the current AI Services setup flow. A newly started tunnel can need a little extra time before its forward proxy is ready.
TrueTest opened up to AI clients this month, and tightened how its Journey Maps track what you're actually looking at.
From an authorized MCP-compatible AI client, you can discover TrueTest applications and tracking environments, search and inspect captured User Sessions, and search, inspect, or request generation of User Journey Maps, bringing captured user behavior and Journey Map workflows into the AI tools you already use while retaining your Katalon project permissions. See the TrueTest MCP tools.
Apply the complete supported User Flow filter set or search criteria to the User Journey Map, so the map and flow list focus on the same result set. Clearing or disabling filter application returns the unfiltered map, and reviewers can analyze the journeys relevant to their current filter without the map and flow list drifting out of sync.
On a flow that already has a linked generated test case, testers can add or remove supported verifications, save the flow, and regenerate the same linked test case, and Apply-to-similar can update eligible matching flows with linked generated test cases after confirmation. Teams can strengthen validations after generation without recreating the test case or losing its identity and repository location.
If July was about getting information out of the platform in a form other people accept, August was about putting control back in your hands and tightening what's underneath it: a real version history instead of silent overwrites, your own taxonomy to manage instead of a ticket to file, an AI Assistant that asks before it acts, Playwright finally reaching the environments the rest of the platform already could, and a security and reliability pass that closes out the review period.