Participant/expert integration and repeatable checks
This is the child-owned frontend evidence. It does not prove publishing, final image generation or independent visual acceptance.
Run the main real-browser interaction checks against the local review build:
python3 -m http.server 8764 --bind 127.0.0.1 --directory outputs/rita-journey-20261002/site
node outputs/rita-journey-20261002/evidence/app/verify-app.cjs
node outputs/rita-journey-20261002/evidence/app/verify-edge-states.cjs
Run the same checks after parent deployment by setting KL_APP_BASE to https://review.clientsflow.hu/komplex-journey-2026-10-02-v1/app/. Both scripts close all their headless browser contexts. They never open a user-facing panel or tab. Customer viewports are only 360/390/430.
verification.json contains the final full-loop receipt: each day’s exact full original module, 4 chooser destinations, reflection edit/save/reload, unsupported/no-file/valid/video replacement/removal/playback/persistence/submission, expert draft/edit/preview/send, participant readback and reviewed status. It also checks all 31 route layouts at the three mobile widths, day switching, complete consultation branches, runtime errors and request methods.
edge-verification.json checks starting-video separation, corrupted MP4 rejection, untouched slider validation, real storage failure without false save, postcall preview without invented booking, a second participant’s isolated reflection/video/draft/send/readback, and live feedback readback in an already-open paired mobile view. The second expert fixture is selected through the participant dropdown and participant=sample-b. Its data is separate, including videos, initial clip and consultation.
source-integrity.json proves 6 byte-identical originals, 6 full rationale passages, 29 task sections and the module-1 introduction. Every day also displays the complete original in a labelled source expansion and offers its exact download. source-map.json maps line ranges and the proposed daily focus. source-map.md explains the distinction between source cadence and new daily UI.
Parent integration still owns all 56 new GPT daily images and actual independent postdeployment mobile section screenshots, content/UX reviews and any Luna fixes. Missing parent-owned images are recorded, and the app shows an honest unavailable-image state. Do not mark that dependency passed from chooser references alone.
The fixture MP4 is an eight-second, nonhuman technical playback sample created locally. It is not a Rita lesson, an exercise demonstration, a participant recording or one of the 83 required GPT images. All fictitious feedback text says it describes an imagined practice situation, not the technical MP4.