Release notes··3 updates

A chat agent that reads your test inbox, PR triage for API reviews, and cache-free runs

3 updates shipped on October 12, 2026.

  1. Highlight

    Featured this release

    The chat agent can read your test inbox

    Chat and PR reviews can look at the emails and SMS a test received, so email failures get a real answer.

    Three new chat tools let the agent search the project inbox, open a specific email – readable body, every link with its anchor text, and attachment info – and list SMS messages. Scoped to a test, they show which messages went to that test's addresses and whether one arrived after the test had already moved on.

    For AI QA testing of sign-up and login flows, the PR review bot uses the same tools to tell a slow delivery or a test-side misread apart from a real app bug when an email test fails.

    More on this: authentication testing, notifications testing and PR testing.

    # Direct link

  2. Update 02 of 03

    PR triage applies to API and GitHub Action reviews

    Reviews started through the API or the GitHub Action follow your PR Triage setting, with an optional per-review override.

    Change reviews started with POST /v1/chat/change-review or the GitHub Action now follow your repository's PR Triage toggle and prompt, the same as mentions and "Start review". A skipped review shows up as a skipped QA.tech / PR Review check with a short summary.

    To override a single review, send an optional triage object: triage.enabled (false to never skip, true to allow a skip even when the toggle is off) and triage.prompt (up to 4,000 characters) to replace the triage prompt for that review only.

    More on this: PR testing and PR and CI/CD testing.

    # Direct link

  3. Update 03 of 03

    Run without the agent cache from the API and MCP

    Runs and reruns started from the API or MCP accept a flag that skips Agent Cache for every test in the run.

    Pass "skipAgentCache": true to POST /v1/run or POST /v1/run/{shortId} (rerun), or to the start_run_with_test_cases, start_run_with_test_plan and rerun_run MCP tools, to run every test in that run without Agent Cache. It's a quick way to check whether a flaky test is a cache problem, without unticking the setting test by test. Omitting it or sending false keeps each test's own setting.

    More on this: MCP, API testing and PR and CI/CD testing.

    # Direct link

Your team moves fast. Can your testing keep up?

QA.tech agents test your product autonomously, so moving fast never means shipping broken. See how it works in a 30‑minute demo.

Book a technical session