Notifications

AutoTestX emails people about things that happen when nobody is watching the app:

What	When	Who, by default
Scheduled runs	A schedule’s suite starts failing, or passes again	The schedule’s own Email on failure list. See Schedules → Failure emails.
Deploy pipeline runs	A suite run started by a pipeline starts failing on an environment, or passes again	Admins and Test Managers of the project
Self-healing fixes to review	New locator fixes are waiting for a review	Everyone who may approve heals: Test Developers, Test Managers, Admins

The last two are set in the Notifications card on the Test Suites page.

Emails go out through the server’s mail settings (Configuration → Email). Without them, the server prints each email in its console instead.

Deploy pipeline runs

When a pipeline starts a suite with an API token (see Run after deployment), AutoTestX tracks whether that suite is passing or failing on each environment and emails when that changes:

It starts failing: Smoke is failing on Staging after a deploy (Web). The email lists the failed tests, why each one failed (the first failed step and its error), a link to each report, and the deploy: source, branch, commit, and a link back to the pipeline run.
It passes again: Smoke is passing again on Staging (Web), with how long and for how many runs it had been failing.

A suite that keeps failing sends one email, not one per deploy. Each environment is tracked on its own: Smoke failing on Staging and passing on Production are two separate stories. A cancelled run changes nothing.

Runs a person starts with Run on the Test Suites page are not emailed, because that person is watching. Scheduled runs use their schedule’s own list.

Example:

Subject: Smoke is failing on Staging after a deploy (Web)

The Smoke suite failed on Staging, in a run started from a deploy pipeline.

Run:     Thu, Oct 8, 10:02 AM UTC
Result:  11 of 12 passed, 1 failed
Deploy:  github-actions · main @ a1b2c3d4e5f6 · #4711
Pipeline: https://github.com/acme/web/actions/runs/4711

Failed test:
  - Apply discount code
    Assert text "Discount applied": Expected text 'Discount applied' but found 'Code expired'
    https://app.autotestx.com/#/runs/run_…

Open the run: https://app.autotestx.com/#/suites/exec_…

You will get one more email when Smoke passes on Staging again - not one for every failing run.

Self-healing fixes to review

When a run self-heals a step, the test keeps passing on a locator nobody has checked yet until someone approves or rejects the fix. The digest makes sure a reviewer hears about it:

At most one email per project per hour (set by the server’s ATX_HEAL_DIGEST_MINUTES), and only when new fixes were found since the last one.
It lists the new fixes, highest risk first: test, step, risk, confidence, whether the run used the fix to keep going or only suggested it, and in how many runs it was seen. Each links to that test’s suggestions.
It ends with how many fixes in total are waiting, and a link to the Self-healing page.
Fixes a self-healing rule applied or rejected on its own are not waiting for anyone, so they are not included.
A fix found in the last two minutes waits for the next digest. Its run may still be handing it to the project’s rules.
When the server starts sending digests for the first time, fixes that were already waiting are not emailed. Only new ones are.

Choosing who is emailed

On Test Suites → Notifications, each kind shows who is emailed now. Edit opens the project’s people with a checkbox each.

Save with people ticked: exactly those people.
Save with nobody ticked: turns that notification off.
Use default: back to the role-based default. The default follows the roles: someone promoted to Test Manager starts receiving pipeline alerts without anyone editing the list.

Changing the lists needs the Manage scheduled runs permission (Test Managers and Admins). Everyone in the project can see who is on them.

People who leave the project stop receiving emails, even if they are still on a list. API tokens never receive email.

Teams and Slack

Email is the only channel today. Posting the same messages to a Microsoft Teams or Slack channel is planned. Until then, most teams forward the alert address to a channel by email (both Teams and Slack can give a channel its own email address).

Troubleshooting

You see	Why	Fix
No email at all	The server has no mail settings	Set `ATX_SMTP_URL` and `ATX_MAIL_FROM`, or read the emails in the server console
No email after a failed deploy	The suite was already failing on that environment	Expected: one email when it starts failing, one when it recovers
No email after a failed run you started	Runs started from the Test Suites page are not emailed	Expected. Use a token from a pipeline, or a schedule.
No heal digest	Nothing new since the last digest, or the fixes were decided by a rule	Check Self-healing → Pending
Heal digest arrives twice from one server	Two server processes share one data folder	Set `ATX_SCHEDULER=off` on the second one

To try these emails step by step, see the exercises (sections A, B and C).