Treseko 1.0.6: QA, AI and traceability updates

A release is best understood by what it lets you do

Treseko 1.0.6 brings together improvements for designing tests, retaining evidence, and understanding build risk. The thread is simple: release decisions should have context, without reconstructing it from a dozen open tabs.

Some of this work accumulated since 1.0.2. The links below take you straight to the relevant feature guide.

Tests teams can repeat and review

API tests declare requests, variables and assertions; conversational tests arrange messages and expectations turn by turn. In both cases, the team can return to the result and its evidence instead of reconstructing what happened from a chat. Case format and execution mode remain separate concepts.

AI to prepare work, with the team in the loop

AI assistance can suggest requirements, stories and test cases. Teams can also configure profiles, models and workflows for different tasks. Suggestions are a starting point: people review criteria, coverage and assumptions before applying them. That cuts repetitive work without treating a model response as an approved decision.

From requirement to result, with the links intact

Traceability connects requirements, stories, cases, executions and bugs. It helps show which need a test covers, where coverage is missing, and which result or defect is associated with it. When a story changes, the team has more context for deciding what to review instead of chasing references across documents.

Reports people can share, with clear limits

Executive and team reports present QA information for different audiences; snapshots preserve a specific view of results. Shared links can expire or be revoked. Availability depends on edition and permissions: sharing a report does not open the whole installation.

Permissions that match people’s responsibilities

Organization, member, role and permission controls help give each person access suited to their work. The aim is to let collaborators see and do what they need in their scope, without using administrator access as a shortcut for every request. Exact options depend on the role and the capabilities enabled in the installation.

Project documentation and inventory, close to the work

The project wiki keeps decisions and procedures in Markdown pages tied to the project; inventory records assets and endpoints that help explain a test environment. This makes context easier to find when someone picks up a case. Neither replaces execution evidence; inventory is not a performance runner, and the wiki should not store secrets.

Workers that receive compatible jobs

Registered workers report their status and capabilities. Treseko uses that information to avoid assigning new work to inactive or revoked workers and to check compatibility with the declared framework and language. A worker executes and returns results; Treseko remains the source of truth for cases, executions and evidence.

MCP integrations with clear permissions and controls

Compatible tools can connect through MCP to read or change information according to enabled functions and permissions. This lets teams integrate assistants without giving them general access to the installation. Available operations depend on scopes and roles; MCP is not a console for arbitrary commands.

Updates with checks and a way back

Treseko prepares and verifies components before activating an update. The new version is confirmed only after services start and pass their checks; if the process fails, a backup and recovery path are available. This reduces the chance of leaving an installation half-updated, while operational backups are still necessary.

Less friction in everyday work

The release brings together improvements to navigation, attachments, evidence and reports, with better attention to smaller screens. The application interface is available in Spanish, English and Portuguese. These changes help teams find what they need sooner and review results in the context where they work.

Security and automation status

Sensitive-data protection, connection validation, and credential and permission management were also strengthened. If Treseko cannot confirm an automation result, it keeps the status as unknown instead of marking it as passed.

Automation is not blind approval

If an execution result cannot be determined, Treseko keeps it unknown instead of marking it as passed.

The linked guides explain each capability, its controls, and its usage requirements.