Release notes··5 updates

    Config descriptions, smarter API keys, fast-agent defaults, and one-time-use data

    5 updates shipped on June 18, 2026.

    1. Highlight

      Featured this release

      One-time-use test data configs

      Test scenarios that need unique values – SSNs, phone numbers, lottery codes – now have a dedicated config type that hands each run a fresh item from the pool.

      Under Project Settings → Configs → Add config, a new "One-Time Use Items" type lets you define a pool of single-use values by typing them in manually or uploading a CSV. Each test run consumes the oldest available item the moment it starts; parallel runs each get a distinct value. The config panel shows an available/used toggle per item so you can track what's left. When the pool runs dry, the run fails with a clear message rather than silently reusing values. You can also create this config type from the chat agent or reference it in test cases via configShortIds in the API.

      # Direct link

    2. Update 02 of 05

      API key management improvements

      See when API keys were last used, and create org-scoped service keys directly from settings.

      The API Keys table in Settings → API now shows a "Last used" column for each key, so you can spot stale or active keys at a glance. Org admins can also now create service keys – choose org-wide or project scope and set read or write permissions – without leaving the dashboard. Personal key management (create and revoke) works the same as before.

      # Direct link

    3. Update 03 of 05

      Config descriptions help the assistant pick the right login

      Add a description to any login config and the assistant will use it to pick the right one automatically – no more duplicate Memory entries.

      Project configs (Settings → Project → Configs) now have an optional Description field, placed directly below the config name. When the assistant needs to decide which login to use for a test, it reads that description alongside the name. A note like "Use for privately-paying patients; don't use for regionally-funded flows" is now enough – you don't need to mirror the same guidance into Assistant Rules. The same context is available to the test agent when it runs, so the right credentials are chosen consistently across both chat and scheduled runs.

      # Direct link

    4. Update 04 of 05

      New tests default to the faster agent

      Tests you create now run with the Fast agent by default; existing tests are unchanged.

      When you add a test case, the agent picker shows "Fast (default)" – you no longer need to select it manually. Any test created before this change keeps whatever agent it was already using. Fast tests are typically quicker to run; switch to Intelligent in the test editor if a particular flow needs deeper reasoning.

      # Direct link

    5. Update 05 of 05

      View code diffs inline in the chat

      When the chat agent surfaces a pull request, you can now expand any file to read its diff without leaving QA.tech.

      Each file row in the PR card is now clickable. Click to expand and see the full diff – additions in green, deletions in red – with hunk headers highlighted. Multiple files can stay open at once. Files without diff data show a "No diff available" notice. Previously, checking what actually changed meant switching over to GitHub.

      # 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.

    Get a demo