Release notes··6 updates

OpenAPI coverage, unattended exploratory reviews, and performance in test plans

6 updates shipped on November 17, 2026.

  1. Highlight

    Featured this release

    Import OpenAPI endpoints into the coverage map

    The coverage map imports API endpoints from OpenAPI and Swagger specs and connects them to calling pages.

    The Coverage page now imports endpoints from OpenAPI 3.x or Swagger 2 specifications found at common paths and refreshes them daily. Endpoints are clustered by tag, shown as diamonds or stars when tested, with dashed outlines for undocumented endpoints.

    Animated lines connect web pages to the endpoints they call. Selecting an endpoint shows where it was observed and which pages call it. API tests receive the endpoint list before execution, so the agent knows which endpoints are available.

    More on this: API testing, API contract testing and API regression testing.

    # Direct link

  2. Highlight

    Featured this release

    Exploratory reviews plan probes and run unattended

    Exploratory reviews plan their probes, execute parallel test batches and run unattended when started by an automation.

    Exploratory reviews use project knowledge and, when enabled, code access to plan probes for state changes, sorts and filters, edge-case inputs, failure paths, narrow screens, keyboard use and other users. They execute parallel batches of short draft tests, applying agentic testing to the planned scope.

    With a time budget, new batches continue until about 80% of the budget is used. Each distinct confirmed bug gets a finding, without the previous 20-finding cap. Automation-started reviews take their scope from the kickoff message and run without stopping for questions.

    More on this: exploratory testing and automated exploratory testing.

    # Direct link

  3. Update 03 of 06

    Four PR review verdicts with named bugs

    PR reviews distinguish confirmed bugs, full verification, partial verification and changes that could not be verified.

    PR reviews now finish with Bugs Found, Fully Verified, Partially Verified or Can't Verify. Bugs Found means a test confirmed a functional bug and the GitHub check fails. Fully Verified and Partially Verified pass the check; Can't Verify is non-blocking.

    Any confirmed bug on a page reached by the tests counts as Bugs Found, including a pre-existing bug, and can block a merge under branch protection. Improvement suggestions do not affect the status. The summary names each confirmed bug, describes the three worst individually and lists any others together.

    More on this: PR testing and the pull request checklist.

    # Direct link

  4. Update 04 of 06

    Show PR risk as a GitHub label

    A PR review rule can label GitHub pull requests with their assessed risk level.

    Add Risk Label on the PR from the rule library as a PR review rule. QA.tech applies qa:risk:low, qa:risk:medium, qa:risk:high or qa:risk:unknown as soon as risk assessment finishes, before test creation.

    A later commit that changes the risk level replaces the old label. QA.tech only adds or removes labels beginning with qa: and creates missing labels when needed. This rule currently works on GitHub.

    More on this: risk-based PR testing and PR testing.

    # Direct link

  5. Update 05 of 06

    Track performance on every test plan run

    Test plan settings can enable performance recording for scheduled, CI, API and manual runs.

    Open a test plan → Settings and turn on Record performance metrics in the Performance Metrics section. Every run of that plan records performance, including scheduled, CI and API runs. Chat can also enable or disable recording when creating or editing a plan.

    Run reviews with performance data include a Performance section identifying vitals outside Google’s good range and the affected pages. It lists up to five of the slowest tests with a short measurement note.

    More on this: test plans and web testing.

    # Direct link

  6. Update 06 of 06

    Manage applications, environments and scenarios through API and MCP

    New API and MCP operations manage application settings, environment access and test scenario folders.

    The API can update application names, colors and session state lifetime. Environment operations can delete environments or create and update custom header rules and HTTP Basic Auth credentials, matching Settings → Applications & Envs. Passwords are write-only. Scenario operations list, create, rename, move and delete test scenario folders.

    MCP now uses one manage_* write tool per entity, such as manage_rules, manage_test_plans and manage_environments, with operation: create | update | delete. The old per-operation tools remain compatible but are hidden from the tool list.

    More on this: MCP for coding agents, API reference and applications and environments.

    # Direct link

Your team moves fast. Can your testing keep up?

Our QA agents test your product autonomously, so moving fast never means shipping broken. Start free or get a 30-minute demo.