The full-scenario demo: one spec, e2e test or video capture - #75
Merged
Conversation
…pture A plain-language laboratory question becomes checked SQL, is refined in conversation, both results are saved as governed Datasets, a table and a grouped-bar Widget are built over them, a Dashboard collects both, Publish to Superset writes the native bundle, the pinned importer brings it in (the spec runs it, so the Bundle ready -> Imported flip is covered end to end), and the finished dashboard renders in Superset. One spec, two Playwright projects: deterministic asserts with no pacing; demo-video is the same steps with dwells, recorded at 1280x720 for the published cut. The demo never diverges from the test, so the recording stays evidence that the product works. scripts/superset-demo-fixture.mjs remains for SQL Lab work (HIV connection + permalink); the demo itself no longer needs Superset UI driving beyond viewing the imported dashboard.
Eight findings from running it against the live stack, each recorded in
the demo issue log:
- the dataset tile's button is TurnNotebook's 'Save to datasets', not the
panel component's aria-label, and the review dialog only closes through
its own Close button — the workspace's control sits behind the backdrop
- the tucked composer is only restored by the floating 'back to [n] · ask'
pill; the toggle is CSS-hidden and the top-right jump control only
scrolls
- on this catalog the writer projects COUNT(*) without the alias its own
structured output promises, and the advisory validation rejects it —
so the demo now pins standing guidance ('Alias aggregate columns
explicitly…') before refining, which is both the fix and a governance
beat worth showing
- the Dashboards library never refetches import receipts on navigation;
a reload is what flips Bundle ready to Imported
The pin bar sat unstyled at the top of the thread — truncated placeholder, orphaned button, and nothing on the page saying what it was for. The owner's read on seeing it: ugly, malformed, no use case visible. Session guidance stays an API-level capability (the harness pins through the gateway and asserts it lands; the demo spec now pins the same way), and the composer can return when it has a designed home. The demo plan also says much less: one-word step cards, captions only where the footage cannot speak for itself.
pmanko
force-pushed
the
feat/full-scenario-demo
branch
from
August 26, 2026 07:38
550520f to
37cd666
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The spec behind the new openclinai.org video, built so the demo can never drift from a working product:
--project=deterministicasserts;--project=demo-videorecords; onlydwell()differs.Issue log from building it (33 findings): staged beside the cut in the harness repo's artifacts/demo-videos/per-dataset/.