# Capture and acceptance ledger — Customer conversations

Recorded 19 September 2026 against the locally repaired Studio, AppEngine and actual chat widget. Separate Cedar employee and anonymous visitor browser sessions were used. The standalone generated HTML example used the same local API/widget build. No production deployment is implied.

## Accepted course actions

| Action | Actual evidence |
| --- | --- |
| New site widget, Agents only, new app user, Finish, private secret handling, reload | 01–06 PNG; 05 secret masked before capture |
| Visitor email/question, queue pickup, reply and follow-up in opposite browser | 07–12 PNG; repeated successful exchange 19 |
| Same-organisation transfer availability after foreign organisation presence changes | 20 PNG/TXT; cross-org-auth-result.json; no actual transfer attempted |
| Saved historical messages | 24 PNG/TXT |
| Assigned thread survives actual Studio reload; reply received and end delivered once | 27–29 PNG; visitor-reload-reply-end.webm |
| Correct generated complete-page instructions | 25 PNG/TXT; generated-embed.html |
| Header/opener save, reload and actual public welcome | 31–32 PNG |
| Agent offline while widget Online | 33 PNG |
| Explicit Offline, Show Form enabled, actual visitor form | 34–35 PNG; offline-settings-readback.json |
| Away submission persists ticket IBQRTCKL and staff reopens matching contents | 41–43 PNG; offline-ticket-readback.json |
| Manual ticket SH78WHBS save/list/reload/readback | 37–39 PNG; manual-ticket-reading-exercise.json |
| Independent reader identifies room, sofa, storage objective and next consultation | manual-ticket-independent-readback.md |
| Offline without form, actual contact anchor and email-link destination | 44–45 PNG; destination is fictional, no email-delivery claim |
| Restore final Online mode and tested away form configuration | 46 PNG |
| Standalone generated example: Alex visitor → queue → agent reply → End | 47–49 PNG; external-widget-exchange.webm |

## Raw recordings

| File | Duration | Frame |
| --- | --- | --- |
| [visitor-reload-reply-end.webm](visitor-reload-reply-end.webm) | 260.52 seconds | 1440 × 1000 |
| [external-widget-exchange.webm](external-widget-exchange.webm) | 43.72 seconds | 1440 × 1000 |

These are actual browser recordings and include natural waits. Files in raw-video are original recorder outputs; the named copies above are the companion packet’s stable references. No edited or narrated final video is claimed.

## Diagnostic evidence, not current lesson outcomes

13 showed an unrelated organisation’s agent before tenant isolation was repaired. 14 recorded the guest reconnect problem. 21 predates lifecycle-notice deduplication. 23 shows missing active controls before assignment recovery. 26 shows the first recovery attempt with an already-expired visitor; 27–29 are the fresh accepted repetition. 30 is an invalid intermediate technical-name edit, discarded before saving the final greeting. 36 is the former general-ticket authentication error, superseded by successful narrow guest intake in 41–43. There is no capture 40: that attempt timed out and is not claimed.

## Regression checks and boundaries

- backend-lock-regression-tests.txt: 23 gateway/session/assignment/lock checks at that stage.
- offline-intake-tests.txt: seven bounded guest-ticket intake cases.
- verify-presence-ui.cjs: six actual UI presence-handler cases.
- verify-recovery-and-notices.cjs: actual restoration/duplicate-notice handlers and metadata filtering checks.
- verify-auth-initialization.cjs: seven actual toolbar initialization cases, added after a separate employee sign-in exposed the transport/authentication race; fresh sign-in/offline/reconnect/reload passed in capture50.
- offline-intake-unauthenticated.json: live unauthenticated intake refused; no extra ticket created.

No actual two-agent transfer, attachment, AI takeover, telephone setup, receiving inbox or transcript delivery was tested. The fictional contact fallback establishes link destination only. Earlier rejected attempts remain as research evidence instead of being presented as successes.

[Course](../../../course-content/appmint-run-customer-conversations.md) · [Production guide](../../../course-content/production/appmint-run-customer-conversations.md) · [Detailed recovery log](PROGRESS.md)