UAT evidence → documentation → demonstrations → user training

This extends Harborline's fictional delivery story. The repository review found source for Scriptor recording/export, BrowserForge browser execution/capture, and Showcase ingestion and verification checks. Their interoperability has not been executed in this documentation task. Harborline's use, approvals, assets, and results below remain hypothetical.

Separate the responsibilities

App	Core purpose and SDLC responsibility	Harborline use	Boundary
ArcoreScriptor	Capture a user's browser workflow into editable steps, screenshots, annotations, and documentation exports	Priya records the coordinator's normal and exception procedures and explains why each decision matters	An observed click is not an assertion or business approval; a recording may contain mistakes
ArcoreBrowserForge	Execute browser workflows and assertions; retain action-linked screenshots, traces, recordings, and interaction captures	Amina repeats approved acceptance scenarios against the candidate and captures evidence	Screenshots alone do not prove the ERP reserved a part; verify authoritative state separately
ArcoreUAT	Organize acceptance cases, execution/evidence, findings, and readiness information	Luis and Amina review scenario outcomes and attach business acceptance to the candidate	Source is now included in the expanded Projects review; production capability and live integrations still need validation
ArcoreUATCLI	Export acceptance evidence and documentation scaffolds	Store a controlled snapshot of the accepted cases and asset references with release records	Client tooling depends on a compatible UAT server
ArcoreShowcase	Normalize source workflows and produce guides, narrated videos, and live demos	Produce coordinator, supervisor, and support training from the accepted workflows	A polished demo does not replace the full test report or acceptance decision
ArcoreVox	Voice transcription/synthesis service	Optional reviewed narration for Showcase training videos	Validate service access and outputs; text guides remain available without voice

The selected source includes Showcase ingestion adapters for UAT, Scriptor, and BrowserForge. That is stronger evidence than a proposed architectural arrow, but it does not prove that today's configured services and schemas work together. One sample recording, one browser session, and one UAT export must pass an ingestion check before the pilot relies on automation.

These are the proposed Harborline handoffs. Require acceptance linkage before publication; ingesting a recording into Showcase does not authorize its training use.

A worked training scenario: the missing seal

1. Capture business knowledge during discovery. Priya uses Scriptor in a seeded training environment to explain how she checks a work order, notices that the required seal is unavailable, contacts procurement, and avoids promising an unsupported arrival time. The initial recording includes a mistaken click. She retains the original as discovery evidence and edits a separate procedural version. The process owner reviews the explanation, not just the screenshot sequence.

2. Turn the recording into acceptance criteria. Amina assigns UAT-03 to R02/R03/R05: an unavailable or uncertain reservation must not become a confirmed commitment. The case specifies starting data, actor/branch, steps, expected visible state, and the authoritative ERP check. It also includes a supported normal commitment and a restricted supervisor exception. The team does not blindly replay every captured click as a valid test.

3. Capture repeatable evidence. BrowserForge executes against the exact candidate environment with synthetic accounts and stable work-order identifiers. Browser assertions check the user-facing state; a separate API check establishes whether the expected reservation exists and whether duplicates were created. The evidence manifest contains application revision/artifact identity, environment, scenario ID, run/session/action IDs, timestamps, and asset hashes. A failed assertion stays failed even when a screenshot looks correct.

4. Preserve the full acceptance record. UAT retains passed, failed, skipped, and interrupted outcomes along with the tester's decision. A retry is a new attempt linked to the original, not a deletion of inconvenient evidence. Showcase imports are teaching material rather than a record of test outcomes, so raw timelines and test reports are preserved separately.

5. Generate a reviewed teaching package. Showcase imports the chosen accepted material, checks required asset availability, and normalizes it into a teaching script. The training lead separates three audiences: coordinators need the decision flow; supervisors need the authority and expiry rules; support needs UNKNOWN-state investigation and escalation. Showcase spec checks can verify state through APIs, but their expected outcomes still need human review. Optional ArcoreVox narration explains why an action is taken. A written transcript and static guide accompany the video.

6. Teach the difficult case as well as success. The normal demo shows a real confirmed reservation in the training environment. The exception demo shows why an unknown outcome must be investigated instead of repeatedly submitted. The supervisor demo shows an independent approver and an expired delegation. The material does not teach trainees that a green notification means the business transaction succeeded.

7. Validate learner performance. A different coordinator follows the guide without author coaching, completes the normal request, identifies the blocked case, and escalates UNKNOWN correctly. Luis accepts the package only if the user can perform the work, not merely play the video. Training feedback becomes a documentation finding or product defect with an owner. If the finding changes functionality, the relevant requirement, build evidence, and acceptance are renewed.

8. Publish and maintain by release. The training owner publishes the approved package to Harborline's controlled SharePoint location and links it from the release record. Each lesson identifies the supported application release, role, prerequisites, owner, and review date. A UI or business-policy change marks affected lessons for review. Regenerated assets remain draft until reviewed; an obsolete lesson is retired or visibly archived rather than silently left beside the current one.

Training deliverables and gates

All IDs below are proposed narrative artifacts, not generated videos or executed UAT results.

Artifact	Audience / owner	Source and release linkage	Acceptance condition
TRAIN-027-01 coordinator quick-start	Coordinators / Priya	Scriptor procedure + accepted normal workflow	A new operator completes the workflow unassisted; labels match the candidate
TRAIN-027-02 exception walkthrough	Supervisors / Luis	UAT override case + BrowserForge evidence	Shows delegated authority, independent approval, expiry, and escalation
TRAIN-027-03 UNKNOWN recovery guide	Support / Grace	R03 and R12 cases + incident lesson	Does not encourage blind retries; verifies remote state and escalation
TRAIN-027-04 narrated demonstration	All pilot users / training lead	Showcase script, captured assets, optional Vox audio	Expected-result checks pass; narration, transcript, and visible actions agree
TRAIN-027-M manifest	Release owner	Links all source revisions, candidate identity, runs, assets, approvals, and publications	No missing required assets or references to a different unreviewed release

Do not assume a screenshot annotation protects the underlying raw file. Use seeded data, inspect final exports including transcripts and video frames, and control raw artifact access and retention separately. Configure BrowserForge storage, and all publication, to Harborline's approved storage and access model.

Framework revision 1.5

Add the following checks to steps 05, 12, 13, 14, and 16:

• Specified in the narrative: requirements identify the learner roles and the normal, exception, and recovery procedures.
• Specified in the narrative: execution evidence includes assertions and authoritative-state checks; recordings and screenshots are supporting evidence.
• Specified in the narrative: business acceptance and the training owner approve release-linked documentation and demonstrations.
• Specified in the narrative: release readiness includes observed user practice, complete source-to-publication manifests, and restricted artifact access.
• Specified in the narrative: outcome reviews include task success after training, support requests, and stale lesson count.
• Real execution pending: validate configured Scriptor → Showcase, BrowserForge → Showcase, and UAT → Showcase imports and required asset retrieval.
• Real execution pending: record and review the actual lessons, test learner completion, and publish approved materials.

The extension uses Harborline's existing 300-hour training/operational-acceptance allocation. It does not assume additional tooling is free: reserve effort for connector validation, asset review, hosting, voice services, and maintenance, then revise the forecast if the allowance is insufficient. The story's financial and outcome numbers remain its original hypothetical estimates, not measurements of these newly added integrations.