Know within minutes whether a deploy broke your Ghost Inspector tests

**Objective:** Find Ghost Inspector tests that passed before the latest successful run of the GitHub workflow `{workflow-name}` (branch `{branch}`, or any if blank) and have failed since.

**Step 1: Find the deploy (GitHub)**

Call `github-get-workflow-run` with `name` set to `{workflow-name}`, `conclusion` success, and `branch` set to `{branch}` if given. Take the most recent run by `updated-date` and treat its `updated-date` as the deploy time. If there is none, stop.

Call `github-get-workflow-run` again for the previous successful run of the same workflow, to bound the before window.

**Step 2: Compare results (Ghost Inspector)**

Call `ghost-inspector-get-test-result` twice: once with `date-created.from` set to the previous deploy and `date-created.to` set to the deploy time (before), and once with `date-created.from` set to the deploy time (after). Group by `test-id` and take the latest result per test in each window.

* **Regressed**: passing before, failing after.
* **Fixed**: failing before, passing after.
* **Still failing**: failing in both.
* **Not run since**: had a result before, none after.

**Step 3: Detail the regressions (Ghost Inspector, then GitHub)**

For each regressed test, call `ghost-inspector-get-test-step` with its after `test-result-id` and `passing` false for the failing step. Then call `github-get-commit` with `commit-date.from` set to the previous deploy and `commit-date.to` set to the deploy time to list what shipped.

**Step 4: Report**

* The deploy: workflow, run number, branch, short sha, time, url.
* Regressed tests first (name, suite, failing step command and target), then Fixed, Still failing and Not run since as counts with names.
* The commits that shipped in the deploy.
* A verdict: whether the deploy looks clean or introduced regressions.

**Rules & Guidelines:**

* `github-get-workflow-run` and `github-get-commit` are on the github server; the ghost-inspector tools are on the ghost-inspector server.
* This prompt is read-only. Do NOT call `ghost-inspector-execute-suite`, `ghost-inspector-execute-test`, `github-rerun-workflow-run` or `github-add-issue`.

**Validation:**

* If no tests ran after the deploy, say the deploy is untested and suggest running the relevant suite; do not run it.
* Compare times in UTC; if either side has no zone, widen the boundary by one hour and say so.