Human-centered AI governance

Learner Journey Labs AI Ecosystem

Explore how specialized AI skills, evidence, quality checks, and human decision-making connect across the Learner Journey Labs ecosystem.

Start with the system map, then open any workflow for more detail.

Module Quality Process Pyramid showing eight layers from Learning Assets through CEO review and SME final release, grouped into Create, Assure, and Release phases.

Detailed explanation of the Learning Assets, Nursing Module Audit, and Whole Module Audit layers, including the checks and controls within each layer.

The whole system in one screen

Read this first. Open the other tabs only when you need the detail behind a stage.

The AI does not create and approve a final module in one step. Specialist ecosystems create assets, formal audits test them, humans authorize movement between layers, and every material change returns through the checks it can affect.

HUMAN STARTS THE JOB1 · Module-governed authoring ecosystemsText · video authoring · video production · standalone page images · embedded H5P/NGN

↻ LOCAL QUALITY LOOP
Trigger: a draft criterion fails → Return to: owning specialist → Repeat: targeted repair and comparison → Exit: authoring-ready candidate

↓ governed candidate

HUMAN DECISION2 · Human checkpoint 1Approves entry into formal Nursing Module Audit or returns the asset to its authoring ecosystem

↓ approved for Nursing Module Audit

HUMAN STARTS WORKFLOW · INTERNAL ROUTING IS AUTOMATIC3 · Nursing Module AuditPrimary production audit, formal scoring, and three remediation batches for text, video, and embedded assessments

↻ Nursing Module Audit LOOP
Trigger: score below 100 or open finding → Return to: correct batch and specialist → Repeat: repair, rebuild, and Nursing Module Audit → Exit: exact module at 100

↓ Nursing Module Audit-verified exact module

HUMAN DECISION4 · Lead human reviewSeparate from Nursing Module Audit and Whole-Module Nursing QA. The Lead returns changes through Nursing Module Audit or authorizes entry into Whole-Module Nursing QA.

↓ Lead approval

HUMAN STARTS ONCE · INTERNAL ROUTING IS AUTOMATIC5 · Independent Whole-Module Nursing QAIndependent final challenge: eight audit domains + four control functions test how the Nursing Module Audit-verified assets work together as one learner experience

↻ Whole-Module Nursing QA LOOP
Trigger: failed criterion, signal, regression, or release check → Return to: exact owner plus local and Nursing Module Audit requalification when affected → Repeat: repair and Whole-Module Nursing QA regression → Exit: Ready

↓ Whole-Module Nursing QA-verified release candidate

FINAL HUMAN DECISION6 · CEO final human reviewSeparate from Whole-Module Nursing QA. The CEO accepts the exact candidate or requests a governed revision.

↻ HUMAN REVISION LOOP
Trigger: the CEO requests a change → Return to: owning skill → Repeat: every affected authoring, Nursing Module Audit, Lead, Whole-Module Nursing QA, regression, and release gate → Exit: the CEO accepts the new exact version

↓ the CEO accepts

OUTPUT7 · Final outputApproved for the intended build, upload, publication, delivery, or archival workflow

Humans define intent, review meaning, resolve uncertainty, and authorize release. Skills perform specialized work, preserve evidence, test quality, and route corrections. “Automatic” means internal routing after a human starts a governed workflow; the system does not run continuously or approve itself.

Start with what you are trying to do

Your goal	Start with	Then it routes to
Create or revise a text page	Nursing L2–L5 Text Page Writer	Local page loop → human checkpoint → Nursing Module Audit
Create or revise a video script	Correct standard or LJL video-script author	Video authoring loop → approved-script handoff → Video production
Produce video scene images	QA Caption-Safe Video Scenes	Segment → describe → QA → generate in batches → inspect → package → Nursing Module Audit
Create a standalone page image	Instructional Image Description Designer	Approve description → generate separate image → inspect → insert → Nursing Module Audit
Create an embedded H5P or NGN	Correct L2–L5 H5P/NGN skill	Local validity → page context → assessment batch → Nursing Module Audit
Create a standalone NGN bank	NGN Question Bank Author	Evidence-first audit → blind re-solve → reconciliation → human release
Create a cumulative MCQ bank	Correct L2–L5 MCQ writer	Evidence-first MCQ audit → adversarial solve → synchronized release
Create a CJMM case study	CJMM authoring layer	Evidence, unfolding-state, validity, set, remediation, package, and human-review loops
Audit a complete module	Nursing Module Audit	Three remediation batches → Nursing Module Audit to 100 → Lead
Run independent final QA	Whole-Module Nursing QA Orchestrator	Eight domains → challenge → remediation → regression → CEO

Why there are two module audits

Audit	Plain-language purpose	What it authorizes
Nursing Module Audit	Primary production audit and remediation authority. It checks each module asset and its integration against the formal rubric.	A score of 100 permits separate Lead review.
Whole-Module Nursing QA	Independent final challenge. It reopens the exact source and tests whether the Nursing Module Audit-verified assets work together as one coherent learner experience.	Ready status permits separate CEO review.

Show boundaries, roles, terms, thresholds, and the loop-reading key

Separate ecosystems connect without sharing the same release path

Ecosystem	Its own release route	When it reconnects with module governance
Standalone NGN bank	NGN author → evidence-first audit → blind re-solve → remediation → package validation → human release	An NGN embedded in a text or video page also enters the page, Nursing Module Audit, Lead, Whole-Module Nursing QA, and CEO routes.
Cumulative MCQ bank	L2–L5 MCQ writers → evidence-first audit → adversarial solve → set remediation → synchronized bank release	It remains outside Nursing Module Audit and Whole-Module Nursing QA unless a separate integration contract explicitly places part of it inside module scope.
CJMM case study	Case authoring → evidence and unfolding-state audits → blind validity review → remediation → package validation → human release	Any module-governed text, video, interaction, or visual changed by the case returns through its owning local, Nursing Module Audit, and Whole-Module Nursing QA gates.

Who starts work and what runs automatically

Moment	Responsibility
Human starts	Selects the job, supplies the governing inputs, and invokes the correct authoring, audit, or orchestrator skill.
Specialist works	Creates or repairs only its assigned asset and runs its local comparison loop.
Orchestrator routes	Starts its internal subskills, validates handoffs, consolidates findings, and sends failures to the exact owner automatically.
Automatic means	Internal checks and routing continue after a human starts the governed workflow. It does not mean continuous operation, self-directed scope changes, or self-approval.
Human decides	Approves movement between governance layers, requests changes, resolves judgment conflicts, or accepts the final exact version.

Six terms used throughout the map

Nursing Module Audit	Nursing Module Audit—the first formal module-level scoring and batch-remediation ecosystem.
Whole-Module Nursing QA	Whole-Module Nursing QA—the independent final-QA ecosystem that begins only after Lead approval.
Governed candidate	An exact version that passed its current layer and is eligible for the next review; it is not necessarily final.
Targeted repair	Change only what failed and preserve every unaffected accepted component.
Regression	Recheck the changed area and every dependent area to prove that a repair did not create a new problem.
Exact version	The specific files and source fingerprint to which the evidence, scores, findings, and approvals belong.

Visual key

Specialist skill	Creates or repairs one specific asset type at one governed cognitive level.
Audit skill	Compares the actual work with fixed evidence and quality criteria.
Batch skill	Reads findings and routes each repair to the correct specialist without replacing that specialist.
Orchestrator	Coordinates multiple skills and internal handoffs after the human starts the larger workflow.
Human decision	Authorizes entry, resolves a conflict, or accepts/rejects an exact version.
Audit or orchestrator	Compares evidence with governed criteria and controls release status.
Batch or file handoff	Routes findings or carries state between separate workflows.
Quality gate	Determines pass, fail, or blocked status.
↻ Closed loop	Returns work to the named earlier check until its exit condition is met.

Quality thresholds are different gates

Threshold	Where it applies	What it means	What it does not mean
90 or higher plus explicit approval	Standalone page-image description	The description is eligible to be sent to image generation.	The generated image or containing page has not passed yet.
9.5/10 or 95	Video segmentation and caption-safe scene-description package	The production plan is ready for downstream image generation.	The actual scene files and video page have not passed yet.
100	Actual learner-facing asset or Nursing Module Audit	Every applicable criterion is supported by exact-version evidence and no governed change is pending.	It does not replace Lead, Whole-Module Nursing QA, or CEO review.
Ready	Whole-Module Nursing QA	Every applicable domain is 100, all findings and signals are closed, regression passed, and both workbooks validate.	It permits CEO review; it is not CEO’s acceptance.

How to read every closed loop

Loop element	Question it answers	Example
Trigger	What caused the loop to reopen?	A criterion failed, evidence is missing, a human requested a change, or regression found a new problem.
Return destination	Which exact skill owns the failed work?	A script defect returns to the video writer; an image defect returns to graphics production.
Repeated work	Which comparisons and dependent checks must run again?	Repair, source comparison, local validation, dependent assessment checks, Nursing Module Audit or Whole-Module Nursing QA regression as applicable.
Exit condition	What proves the cycle may stop?	The named gate passes on the new exact version with no unresolved work.

Show the longer beginner walkthrough

The first authoring cycle

Start here before looking at any skill names. This first cycle creates a candidate; it does not create the final release.

AI produces a draft. It does not approve its own work. Separate governed skills compare the work with fixed standards, route exact repairs, and repeat the review until the release gate is met.

1 · You define the jobProvide the source, exact objective, Bloom level, asset type, constraints, and existing audit findings.

→

2 · A specialist createsThe correct level-specific writer or designer builds only the assigned text page, video, NGN, MCQ, assessment, or visual.

→

3 · The authoring quality loop comparesThe draft is checked against the source, objective, Bloom contract, clinical and language standards, required worksheet, and output contract.

→

4 · Failed criteria returnEach concrete finding is routed to the exact skill that owns the failed content. Correct work is preserved.

→

5 · The first governed candidate is assembledThe exact asset, synchronized worksheet, validators, render evidence, fingerprints, findings, and limitations are packaged for human review.

→

6 · The human reviews the first governed candidateApprove it to enter module-level QA or return it to the owning authoring loop. This is not the final release decision.

What a person experiences from beginning to final output

The system is a series of governed handoffs, not one AI prompt.

1 · A person defines the assignmentWhat must be created · source of truth · learning objective · Bloom level · constraints · protected content · required files

↓

2 · The correct specialist ecosystem creates and self-checksText, video, embedded H5P/NGN, standalone NGN, MCQ, or image production follows its own standards and local repair loop

↓

3 · The first human reviews the governed candidateReturns it to the authoring loop or approves the complete current module to enter formal Nursing Module Audit

↓

4 · Nursing Module Audit evaluates the actual moduleCreates evaluation and remediation workbooks; no score of 100 is allowed while a change is pending

5 · Nursing Module Audit REMEDIATION LOOP: Trigger: any Nursing Module Audit score below 100 or open finding. Return to: the matching text, video, or assessment batch and exact owner. Repeat: atomic repair, verification, rebuild, and Nursing Module Audit on the new exact version. Exit: the exact module reaches 100 with no open work.

↓

6 · The Lead reviews the Nursing Module Audit-verified versionReturns changes through the applicable batch and Nursing Module Audit loop or authorizes entry into independent Whole-Module Nursing QA

↓

7 · Whole-Module Nursing QA independently challenges the moduleEight audit-domain subskills test asset learning, interaction validity, options, assessment set, visuals, clinical language, progression, and cross-asset redundancy; four control subskills challenge scores, plan remediation, execute verification, and validate release

8 · Whole-Module Nursing QA REMEDIATION LOOP: Trigger: any failed criterion, unresolved signal, regression, workbook, or release check. Return to: the exact owner plus affected local and Nursing Module Audit requalification. Repeat: repair and replay every dependent Whole-Module Nursing QA domain. Exit: all applicable domains equal 100 and release validation passes.

↓

9 · The CEO reviews the exact release candidateAccepts it for the intended downstream use or requests a governed change

CHANGE REQUESTED

10A · HUMAN ITERATION LOOP: Trigger: the CEO requests a change. Return to: the exact owning skill. Repeat: preserve unaffected work and reopen every affected authoring, Nursing Module Audit, Lead, Whole-Module Nursing QA, regression, and release gate. Exit: the CEO accepts the new exact candidate.

ACCEPTED

10B · Final outputThe exact CEO-accepted version may be built, uploaded, published, delivered, or archived. Any later edit creates a new governed version.

Why there are many skills

Specialization	One skill owns one type of reasoning, Bloom level, asset, or production job. Narrow ownership makes failures traceable.
Independent judgment	The skill that creates an asset is not the only skill deciding whether it is acceptable.
Controlled repair	Findings return to the specialist capable of making the smallest valid change without damaging correct work.
Cross-asset control	A page can be excellent alone and still fail the module because it duplicates, contradicts, or disrupts progression elsewhere.

The five nested quality loops

Whole-Module Nursing QA loopEight domains → challenge → synthesize → remediate → regress → validate → rerun.Exit: CEO-review candidate

Nursing Module Audit loopNursing Module Audit → three batch routes → specialist repair → rebuild → Nursing Module Audit again.Exit: Lead-review candidate

Batch loopRead workbooks → separate findings by asset and owner → call specialist → verify repaired asset.Exit: rebuilt master

Local asset loopPlan → draft → compare → targeted repair → resynchronize → revalidate.Exit: authoring handoff passed

Production loopApproved content → segment/describe/generate → QA → regenerate only rejected components.Exit: production package passed

Use the tabs as a map of the whole system

Tab	What it answers
All granular steps	Shows every granular step from assignment through the authoring loop, first human review, Nursing Module Audit and three batch-remediation routes, Lead review, independent Whole-Module Nursing QA, remediation and regression, and CEO’s final decision.
Human checkpoints	Shows where humans review, what authority each human has, what can be returned, and why early approval is not final release.
Operating model	Explains the nested loops, handoffs, evidence records, and why correct components are preserved during targeted repair.
Roles + glossary	Distinguishes creator, evaluator, remediator, validator, human approver, the states an asset moves through, and the system’s specialized terms.
Nursing Module Audit	Shows the first formal module-level scoring ecosystem and its three batch-remediation loops. It ends with a Nursing Module Audit-verified module for separate Lead review.
Whole-Module Nursing QA · 12 subskills	Shows the independent final-QA ecosystem that begins only after Lead approval, including all 12 audit and control subskills, scoring protections, remediation, regression, workbook outputs, and release validation before separate CEO review.
Text	Shows L2–L5 page writing, page interaction and image dependencies, targeted revision, formal Nursing Module Audit, and re-audit to 100.
Video	Shows script authoring, page-level checks, SENSE or governed NURSE, embedded interactions, segmentation, scene descriptions, graphics, and video batch remediation.
Video production	Shows the separate post-script chain: segmentation, source reconciliation, scene descriptions, description QA, batch image generation, per-image QA, package assembly, and return to module governance.
Images	Shows the standalone text-page image description, approval, generation, inspection, regeneration, insertion, and page re-audit loop. Video scenes remain in Video production.
NGN evidence	Shows NGN authoring, evidence mapping, blind re-solving, reconciliation, and the different routes for standalone banks versus embedded interactions.
MCQ evidence	Shows L2–L5 MCQ authoring, evidence-first and adversarial verification, set-level repair, and the separate cumulative-bank release boundary.
CJMM / case studies	Shows the separate case-study authoring, evidence, validity, remediation, regression, and release ecosystem.
All skills	Provides the complete people-facing directory: when each skill runs, what it owns, what it returns, and which workflow receives its output next.

The six questions a person should ask at every handoff

Question	Why it matters
What exact version is being reviewed?	Scores and approvals belong only to the files and source fingerprint that were actually checked.
What gate has passed?	A local writer pass, Nursing Module Audit 100, Lead approval, Whole-Module Nursing QA 100, and CEO acceptance authorize different things.
What can still fail later?	A strong asset can still fail cross-asset progression, redundancy, terminology, assessment distribution, production, or human-fit review.
What evidence proves the result?	The current asset, worksheet, source map, workbooks, validator results, render inspection, manifest, and finding-closure record make the decision auditable.
Who owns the next action?	The human starts a separate workflow when required; the orchestrator or batch routes internal work to the correct specialist.
If this changes, which gates reopen?	The change-impact route prevents both under-reviewing a consequential edit and needlessly redoing unaffected work.

Core idea: Step 6 is the first human checkpoint. The approved candidate still must pass Nursing Module Audit and batch remediation, Lead review, independent Whole-Module Nursing QA and remediation, and CEO's final review before final output.

What a skill is and what happens inside one

Worked example: Nursing Level 4 Text Page Writer

A skill is a governed work contract, not a single prompt. It combines a defined job, authority rules, required inputs, internal reasoning steps, reference standards, templates, scripts, validators, quality gates, failure behavior, exact outputs, and handoff rules.

Technology inside a skill

Component	What it does
Trigger and task mode	Determines whether the run is drafting, evaluating, revising, making one eligible targeted repair, or checking finalization.
Authority hierarchy	Resolves conflicts among fixed objectives, the human's newest decision, exemplars, the worksheet standard, clinical safety, scope, and progression.
References and exemplars	Supply the exact cognitive, clinical, language, architecture, and quality standards used for comparison. They calibrate quality without becoming copied content.
Structured working record	Stores source traceability, planning decisions, comparison evidence, revisions, unresolved gaps, and the final readiness decision in a machine-checkable form.
Builder and template	Turns the structured record into the required staff-facing worksheet or production artifact using the governed layout.
Validators	Check required fields, exact versions, objective traceability, prohibited shortcuts, artifact integrity, synchronization, and release preconditions.
Render inspection	Checks what a person will actually see: clipping, overflow, blank fields, broken tables, awkward page breaks, and readability.
Output contract	Defines the exact deliverables, their order, required links, evidence, status wording, and which downstream ecosystem receives them.
Blocker behavior	Stops release when required proof cannot be generated or validated instead of silently substituting an incomplete result.

Human-centered design inside a skill

Human need	How the system protects it
Intent stays human-owned	The person defines the objective, audience, constraints, protected content, and what success means.
Decisions remain reviewable	The worksheet shows why content was selected, how it maps to the objective, what comparisons were made, and what changed.
Accepted work is preserved	Repairs target the smallest valid scope and name content that must not change.
Uncertainty stays visible	Missing evidence, clinical uncertainty, policy dependence, limitations, and unresolved findings remain explicit instead of being hidden behind fluent prose.
Quality is inspectable	Humans receive the learner-facing artifact plus planning evidence, validator results, render evidence, and a readiness decision.
Approval is separate from creation	A writer may declare authoring readiness, but independent audits and authorized humans control movement and final release.

INSIDE THE L4 TEXT-PAGE SKILL · ONE COMPLETE RUN

HUMAN INPUT1 · Define the teaching jobOne exact MLO, enabling objective, or Level 4 page goal is enough to start. Optional source files, module context, neighboring pages, audit findings, guardrails, and protected content are added when available.

↓ the skill accepts the job without requiring a content draft

2 · Select the task mode and lock the source hierarchyDraft · Evaluate · Revise · one eligible low-severity Targeted Repair · Finalization check. Fixed objectives and the human's newest accepted decisions outrank exemplars and internal planning choices.

↓

3 · Create the structured thinking-proof recordLock ADN prelicensure RN context · exact objective and source inventory · required-content families · guardrails · source-to-page traceability · preservation rules · unresolved evidence

↓

4 · Select at least two complete Level 4 text-page exemplarsChoose one topic or clinical-content match and one architecture, reasoning, or voice match. Record the distinct strength transferred from each and what topic-specific content must not be copied. Video, H5P, NGN, slide, and scene-script assets are excluded.

↓

5 · Plan the reasoning before writingCentral clinical question · foundational dependencies · one-sentence unfolding pattern story · concept-dependency map · clinically logical page order · Level 4 boundary · RN scope · progression from neighboring pages

↓

6 · Build every major Level 4 pattern in the worksheetRelated cue cluster · client-specific comparison point · what changed or no longer fits · concern level · possible safety threat · supported RN response category · mixed-evidence interpretation · single-cue subtraction · diagnostic-certainty control · decision ownership · evidence-specific collaboration

↓

7 · Test two representative content areas before the full draftEach must show the complete relationship chain, require the whole cue cluster, use a meaningful comparison anchor, resist one-cue answering, and remain within Level 4 rather than collapsing to recall/application or drifting into Level 5 strategy judgment.

↓

8 · Write the complete first draft internallyConnected learner-facing prose · purposeful headings · no planning labels · no worksheet language · no bullets, tables, checklists, pseudo-lists, or repeated formulaic mini-cases

↓

9 · Compare the first draft with the objective, sources, standards, and selected exemplarsClinical opening · unfolding story · section dependency · explanation depth · Bloom chain · cluster necessity · comparison quality · mixed evidence · scope · certainty · progression · prose-only compliance · accuracy · ending

↻ FULL-PAGE REVISION LOOP
Trigger: any comparison dimension is Partial, Missing, or Repair → Return to: the earliest failed planning or authoring decision → Repeat: revise architecture, sections, analytical story, explanations, diagnostic certainty, and prose in at least three substantive dimensions for Draft or Revise mode → Exit: the second comparison proves the complete exact page passes every authoring gate

↓

10 · Synchronize the worksheet to the exact final candidateReplace planning language with final evidence · map every objective and source item one to one · identify substantive first-draft-to-final revisions · quote or precisely locate positive evidence · record unresolved gaps and readiness

↓

11 · Run deterministic content validationValidate the structured worksheet against both the exact internal first draft and the exact final page. Correct missing fields, stale evidence, unsupported claims, wrong exemplars, role confusion, scope failures, and synchronization defects.

↻ DATA VALIDATION LOOP
Trigger: the structured record or page fails a deterministic rule → Return to: the exact failed worksheet field or learner-facing passage → Repeat: correct, resynchronize, and rerun the validator against both versions → Exit: the pre-document validator passes

↓

12 · Build the staff-facing DOCX worksheetThe governed builder converts the validated structured record into the task-specific completed worksheet. The static blank is only a visual reference and is never returned as the completed artifact.

↓

13 · Render and visually inspect every worksheet pageOpen the rendered pages and check clipping, overflow, broken tables, blank required fields, boundary-hugging text, awkward page breaks, readability, and document integrity.

↻ DOCUMENT QA LOOP
Trigger: any rendered worksheet page is visually or structurally defective → Return to: the structured record, builder input, or document layout owner → Repeat: correct, rebuild, rerender, and reinspect every affected page → Exit: every worksheet page is readable and complete

↓

14 · Run the final release validatorUse the render directory, completed worksheet DOCX, exact first draft, and exact final page together. Verify synchronization, render evidence, DOCX integrity, required fields, final status, and downloadable-link contract.

Did the skill's final release gate pass?

Return to the exact failed internal ownerDo not release the page alone and do not substitute a summary for the missing worksheet or proof.

↻ BLOCKED OR REPAIR LOOP
Trigger: the final validator, render evidence, DOCX integrity, or working-link contract fails → Return to: the owning internal step or explicit human blocker → Repeat: correct and replay every affected check while preserving passed work → Exit: the exact package passes or a concrete blocker is reported without releasing incomplete content

YES

15 · Return the governed authoring packageFirst: working completed worksheet link · Then: complete final learner-facing page when the task mode requires it · Then: exemplar comparison and evidence-based readiness decision. The internal first draft is never delivered.

What travels through the skill

State	Artifact carried forward	Why it matters to a person
Human request	Objective or page goal, audience, constraints, optional sources, context, protected content, and accepted decisions.	The system starts from human intent rather than inventing the assignment.
Planning state	Structured worksheet record with source inventory, traceability, reasoning map, exemplars, preservation rules, and known gaps.	The invisible reasoning becomes inspectable and correctable.
Draft state	One complete internal first draft bound to the planning record.	The system has something concrete to compare; fluency alone is not treated as quality.
Revision state	Comparison findings, named first-draft defects, substantive revisions, and a new exact candidate.	A person can see that revision occurred and why the page changed.
Evidence state	Synchronized worksheet, exact first and final versions, validator results, render evidence, DOCX, fingerprint, readiness, and unresolved gaps.	The decision is tied to evidence for one exact version rather than a generic assurance.
Authoring handoff	Working worksheet link, final candidate page, exemplar comparison, and readiness decision.	The human receives both the learner experience and the staff-facing proof behind it.

Who each output is for

Output type	Examples	Primary audience and purpose
Learner-facing content	Text page · video page · H5P · embedded NGN · standalone NGN · MCQ · instructional image	Learners. Delivers the approved instructional experience without exposing internal planning or QA language.
Human-facing QA evidence	Thinking-proof worksheet · evaluation workbook · remediation workbook · exemplar comparison · readiness explanation	Authors, reviewers, Leads, SMEs, and the CEO. Makes reasoning, findings, required changes, preservation rules, and approval status understandable.
Machine-checkable records	Structured worksheet record · source fingerprint · normalized manifest · finding ledger · validator result · exact-version hash	Validators, orchestrators, and downstream skills. Proves identity, completeness, traceability, synchronization, and gate status without replacing human judgment.
Production files	Rendered DOCX · segmented script · image-description document · separate image files · title image · one approved cast thumbnail with the full cast facing the viewer · manifest	Production teams and build systems. Carries approved content into the actual learner-facing implementation.
Final release package	Exact approved module · accepted assets · validated Nursing Module Audit and Whole-Module Nursing QA evidence · remediation closure · version record	The CEO and the downstream build, upload, publication, delivery, or archival workflow. Identifies the only version authorized for final use.

OUTSIDE THE SKILL · INDEPENDENT GOVERNANCE BEGINS

HUMAN CHECKPOINT 116 · Review the authoring-ready candidateThe human checks instructional intent, practical fit, requested format, protected content, and whether the candidate may enter formal module audit. The human may return it to any internal L4 skill loop.

↓ approved for Nursing Module Audit

17 · Nursing Module Audit independently scores the actual page in the moduleNursing Module Audit does not inherit the writer's readiness verdict. It reopens the learner-facing source and evaluates objective coverage, Bloom, clinical accuracy, scope, language, progression, redundancy, embedded assessments, visuals, production, and exact-version evidence.

↻ EXTERNAL RE-ENTRY LOOP
Trigger: Nursing Module Audit records a text-page finding or score below 100 → Return to: QA Batch for Text Pages, then this exact L4 skill in Revise mode—or Targeted Repair only for one eligible isolated language substitution → Repeat: the full applicable worksheet, comparison, validation, render, Nursing Module Audit rebuild, and regression checks → Exit: the exact page and module reach 100 with every applicable finding closed

↓ exact Nursing Module Audit-verified module

18 · Lead review remains separateThe Lead accepts the Nursing Module Audit-verified exact version for independent Whole-Module Nursing QA or returns a governed change through the owning skill and Nursing Module Audit.

↓ Lead approval

19 · Whole-Module Nursing QA independently challenges the complete learner experienceA new source index and manifest bind the exact version. Eight domains, a source-blind score challenge, root-cause synthesis, remediation, dependency regression, workbook compilation, and release validation run without replacing the Nursing Module Audit score.

↓ Whole-Module Nursing QA Ready

20 · The CEO makes the final human decisionThe CEO accepts the exact release candidate or requests a change. Any change returns to the earliest affected owner and reopens every dependent gate before a new exact version returns.

↓ accepted exact version

21 · Final outputThe approved version may enter build, upload, publication, delivery, or archival workflow. A later edit creates a new governed version rather than inheriting the old approvals.

Worked-example boundary: The Level 4 writer owns authoring evidence and readiness; Nursing Module Audit owns formal module scoring; the Lead owns entry into Whole-Module Nursing QA; Whole-Module Nursing QA owns the independent whole-module score and Ready status; The CEO owns final acceptance.

Complete governed path from request to CEO-approved output

The asset passes three human checkpoints and two distinct module-level QA ecosystems. No earlier checkpoint is the final release.

PHASE A · AUTHORING ECOSYSTEM AND FIRST HUMAN REVIEW

HUMAN · Step 1Define the assignment and authorityExact objective or page goal · Bloom level · source files · asset type · intended audience · output format · examples · constraints · preserve/do-not-touch instructions · known audit findings

↓

ROUTER · Step 2Select the correct ecosystem, level, framework, task mode, and package scopeText, video, embedded H5P/NGN, standalone NGN, MCQ, or visual · L2–L5 · SENSE or explicit LJL NURSE · Plan, Author, Evaluate, Revise, Targeted Repair, or Finalize · full package, learner-facing only, or partial records

↓

SPECIALIST · Step 3Lock the authoritative inputsFreeze objective wording, source boundaries, clinical evidence, mappings, protected content, filenames, placement, production requirements, and the exact starting version

↓

SPECIALIST · Step 4Build the required planning and reasoning recordComplete the task-specific thinking-proof worksheet, evidence map, content profile, blueprint, canonical state, or production manifest before learner-facing drafting when the governing skill requires it

↓ structural gate

VALIDATOR · Step 5Validate, build, render, and visually inspect planning artifactsRequired fields · exact source rows · traceability · hashes · document integrity · page rendering · clipping · overflow · blank fields · synchronization

↓

SPECIALIST · Step 6Create the provisional learner-facing assetThe correct level-specific writer, designer, bank author, segmenter, description designer, or image skill produces only its assigned asset and preserves all locked boundaries

↓

SPECIALIST QUALITY LOOP · Step 7Compare the provisional version with every governing standardObjective · source evidence · Bloom contract · nursing language · clinical accuracy and scope · exemplar-calibrated structure · accessibility · neighboring assets · assessment independence · production rules

↓

SPECIALIST QUALITY LOOP · Step 8Run semantic and adversarial checksEasiest legitimate answer path · alternate interpretation or answer · strongest competitor · source echo · scenario necessity · read-aloud quality · clinical ambiguity · visual teaching purpose · regression risk

↓

Does the provisional asset pass every applicable authoring gate?

Step 9Record exact failed criteriaFinding · source evidence · owner · required repair · content to preserve · acceptance test · regression scope

Step 10Revise upstream firstUpdate the worksheet, blueprint, canonical state, description, or source-controlled component before regenerating downstream outputs

AUTHORING LOOP: Trigger: the authoring handoff fails. Return to: the failed upstream worksheet, blueprint, canonical state, description, or source-controlled component. Repeat: regenerate, review, synchronize, validate, and rerender the affected asset. Exit: Steps 7–10 pass on one exact version.

YES

Step 11Create the first governed candidate packageExact final asset · required worksheet or workbooks · evidence record · validator result · render result · fingerprints · limitations · unresolved nonblocking notes

↓

HUMAN CHECKPOINT 1 · Step 12Human reviews the first governed candidateConfirms instructional intent, practical fit, requested format, protected content, and whether the candidate may enter formal module-level QA

FIRST HUMAN LOOP: Trigger: the first human requests a change. Return to: the exact authoring owner. Repeat: classified repair plus every affected authoring and validation gate. Exit: the human approves the new governed candidate for Nursing Module Audit—not final release.

PHASE B · NURSING MODULE AUDIT AND BATCH REMEDIATION

HUMAN · Step 13Submit the complete current module packageContent master · exact objectives · text pages · video pages · embedded H5Ps/approved NGNs · visuals · keys and feedback · sources · authoring handoffs when available · preservation instructions

↓

NURSING MODULE AUDIT · Step 14Run source-evidence preflight and freeze the audited versionExtract every supplied source · calculate fingerprints · establish module scope · confirm authoritative sequence · bind assets and interactions to exact objectives and source locations

↓

NURSING MODULE AUDIT · Step 15Audit every governed learner-facing assetText teaching · video teaching · H5P/NGN validity · Bloom · clinical accuracy · nursing scope and language · visuals and production · progression · redundancy · cross-modal contradictions · readiness

↓

NURSING MODULE AUDIT · Step 16Calculate finding-aware scores and release statusEvidence-earned scores · severity caps · no 100 with an unresolved finding · missing evidence left blank rather than converted to zero · exact-version score authority

↓ file handoff

NURSING MODULE AUDIT · Step 17Produce the synchronized workbook pairStrict Exemplar-Calibrated Evaluation workbook · Asset-by-Asset Remediation To-Do workbook · findings, evidence, owners, dependencies, preservation rules, acceptance criteria, and content/H5P task lists

↓ human starts all applicable remediation routes

Step 18AQA Batch for Text PagesRoutes text, worksheet, embedded interaction, terminology, and page-visual findings to the correct L2–L5 writer or supporting skill

Step 18BBatch QA VideoRoutes script, page, H5P/NGN, SENSE/NURSE, segmentation, scene-description, graphic, and package findings to the correct owner

Step 18CQA Batch for AssessmentsRoutes embedded H5P and NGN mapping, Bloom, cue, response, key, feedback, accessibility, and set findings to the correct designer

↓

BATCH + SPECIALISTS · Step 19Apply atomic repairs and preserve accepted workOne finding per repair task · exact source location · smallest valid change · protected images and formatting unchanged unless the finding requires modification · no accepted graphic regenerated without a documented failure

↓

BATCH + SPECIALISTS · Step 20Rerun every affected local gateWriter/designer worksheet · language requalification · clinical/source review · Bloom check · key integrity · segmentation or production QA · rendering · validators · exact-version synchronization

↓

BATCH · Step 21Rebuild the complete content masterInsert only verified repairs · preserve authoritative order · reconcile objectives, assets, interactions, visuals, workbooks, and exact-version identifiers

MODULE AUDIT LOOP · Step 22: Trigger: any score below 100, open finding, changed asset, or workbook defect. Return to: the appropriate batch and exact owning specialist. Repeat: repair, verification, rebuild, all affected dependency checks, and Steps 14–22. Exit: the current exact module reaches 100, no change is pending, and both workbooks validate.

↓

HUMAN CHECKPOINT 2 · LEAD · Step 23Lead reviews the Nursing Module Audit-verified moduleChecks the learner experience, instructional intent, clinical and assessment fit, production presentation, preservation requirements, and whether the module is ready for independent Whole-Module Nursing QA

LEAD REVIEW LOOP: Trigger: the Lead identifies a change. Return to: the appropriate text, video, or assessment batch and exact owner. Repeat: affected local gates, rebuild, and Nursing Module Audit to 100. Exit: the Lead approves the new exact version for independent Whole-Module Nursing QA—not final release.

PHASE C · Whole-Module Nursing QA BEGINS AFTER LEAD APPROVAL · 12 SUBSKILLS

WHOLE-MODULE NURSING QA · Step 24Build a new independent source index and normalized manifestExact source-set fingerprint · module identity · authoritative sequence · stable asset, interaction, visual, and concept IDs · objective allocations · exact-version hashes · disposition for every indexed record

↓

Whole-Module Nursing QA SUBSKILL 1 · Step 25AAsset LearningTeaching quality and learner task

Whole-Module Nursing QA SUBSKILL 2 · Step 25BH5P/NGN ValidityIndividual interaction validity

Whole-Module Nursing QA SUBSKILL 3 · Step 25COption ValidityL3–L5 response integrity

Whole-Module Nursing QA SUBSKILL 4 · Step 25DAssessment SetDistribution and concentration

Whole-Module Nursing QA SUBSKILL 5 · Step 25EVisual ProductionVisual evidence and readiness

Whole-Module Nursing QA SUBSKILL 6 · Step 25FClinical LanguageAccuracy, safety, scope, terminology

Whole-Module Nursing QA SUBSKILL 7 · Step 25GProgressionPrior learning and cognitive growth

Whole-Module Nursing QA SUBSKILL 8 · Step 25HCross-Asset RedundancyRepeated teaching and streamlining

↓

WHOLE-MODULE NURSING QA · Step 26Validate every domain handoff against the same manifestExact coverage · no missing or duplicate rows · matching fingerprints · source-specific evidence · one primary score owner · no recycled or generic evidence

↓ independent challenge

Whole-Module Nursing QA SUBSKILL 9 · Step 27Run the source-blind semantic score challengeHide proposed scores, reopen the exact source, rejudge every criterion, test boundary conditions and counterevidence, reconcile discrepancies, and allow only challenge-verified scores into the ledger

↓

WHOLE-MODULE NURSING QA · Step 28Synthesize findings across domainsMerge root causes · prevent duplicate deductions · reconcile contradictions · build concept, progression, assessment, and change-impact graphs · order dependencies · define atomic remediation and regression scope

↓

Whole-Module Nursing QA SUBSKILL 12 · Step 29Compile the Whole-Module Nursing QA evaluation and remediation workbooksEight-domain scorecard · plain-language change summaries · finding ledger · remediation plan · assessment QA handoff · content and H5P task lists · release-validation status

↓

Whole-Module Nursing QA SUBSKILLS 10–11 · Step 30Plan and execute dependency-ordered repairsRoute each atomic finding to the owning text, video, assessment, visual, language, or production skill; preserve correct content; update the exact version; and capture completion evidence

↓

Whole-Module Nursing QA SUBSKILL 12 · Step 31Replay affected checks and run release validationOwning domain · dependent interactions · clinical contradictions · progression · redundancy · distribution · source fingerprints · workbook synchronization · final score recomputation · read-only validation

WHOLE-MODULE NURSING QA LOOP · Step 32: Trigger: any domain score below 100, unresolved signal, ledger mismatch, workbook defect, or failed release check. Return to: the exact owner plus required local/Nursing Module Audit requalification. Repeat: Steps 24–31 and dependent-domain regression. Exit: all eight applicable domains equal 100 and final read-only release validation passes.

PHASE D · Whole-Module Nursing QA ENDS · SEPARATE CEO FINAL REVIEW

SYSTEM OUTPUT · Step 33Create the Whole-Module Nursing QA-verified release candidateExact module master · final supporting assets · validated Nursing Module Audit and Whole-Module Nursing QA workbooks · completed remediation evidence · version record · package manifest · limitations or explicit blocker status

↓

HUMAN CHECKPOINT 3 · CEO · Step 34The CEO performs final human reviewConfirms the complete learner experience, clinical and instructional intent, quality expectations, protected content, formatting, production usability, stakeholder needs, and readiness for the intended downstream use

Does the CEO accept the exact release candidate?

NO · Step 35

Classify CEO's feedback and determine impactPreference, formatting, instructional, clinical, objective, assessment, terminology, visual, production, or scope change

CEO ITERATION LOOP: Trigger: the CEO requests a change. Return to: the exact owning skill. Repeat: preserve unaffected work and reopen every affected authoring, Nursing Module Audit, batch, Lead, Whole-Module Nursing QA, regression, and release-validation gate. Exit: a new exact candidate is accepted by the CEO.

YES · Step 36

CEO-ACCEPTED FINAL OUTPUTThe exact version is approved for the intended build, upload, publication, delivery, or archival workflow. Any later change creates a new governed version and reopens the gates required by its impact.

What each checkpoint authorizes

Checkpoint	What has passed	What the approval allows	What it does not mean
Human checkpoint 1	Authoring and local validation	The candidate may enter Nursing Module Audit.	Not module verified and not final.
Nursing Module Audit	Formal Nursing Module Audit score 100 and finding closure	The exact module may be presented to the Lead.	Not independently verified by Whole-Module Nursing QA.
Lead checkpoint	Nursing Module Audit-verified module plus Lead judgment	The module may enter independent Whole-Module Nursing QA.	Not CEO-approved and not final.
Whole-Module Nursing QA	Eight domains at 100, remediation complete, release validation passed	The exact release candidate may be presented to the CEO.	Technical readiness is not final human acceptance.
CEO checkpoint	All governed QA plus final human judgment	The exact version becomes the accepted final output.	Later revisions are new governed versions.

Comparison of specialized skill responsibilities and human authority. Skills create, compare, audit, route, repair, validate, and package; humans define intent, approve entry, resolve uncertainty, and authorize release.

The three human quality checkpoints sit outside the audit systems

Each human makes a separate judgment between governed ecosystems. Only the CEO accepts the final output.

Lead review is not Whole-Module Nursing QA. CEO review is not Whole-Module Nursing QA. The Lead decides whether a Nursing Module Audit-verified module may enter Whole-Module Nursing QA. Whole-Module Nursing QA independently creates a release candidate. The CEO then decides whether that exact candidate becomes final.

HUMAN CHECKPOINT 1 · Initial governed-candidate reviewReviews the output of the authoring ecosystem and either returns it to the owning skill or approves it to enter Nursing Module Audit

↓ approved for formal module audit

NURSING MODULE AUDIT + THREE REMEDIATION BATCHESAudit → evaluation and remediation workbooks → QA Batch for Text Pages + Batch QA Video + QA Batch for Assessments → specialist repair → rebuilt master → Nursing Module Audit again until 100

↓ exact module version reaches Nursing Module Audit 100

HUMAN CHECKPOINT 2 · Lead reviewReviews the Nursing Module Audit-verified module and either returns it through the applicable batch and Nursing Module Audit loop or approves it for independent Whole-Module Nursing QA

↓ Lead-approved exact version

WHOLE-MODULE NURSING QA + Whole-Module Nursing QA REMEDIATIONEight independent domains → semantic score challenge → cross-domain synthesis → workbooks → dependency-ordered remediation → regression replay → release validation until all applicable domains reach 100

↓ Whole-Module Nursing QA-verified release candidate

HUMAN CHECKPOINT 3 · CEO final reviewReviews the entire release candidate and either returns exact feedback through every affected governance layer or accepts the exact version as final

↓ the CEO accepts

CEO-ACCEPTED FINAL OUTPUTApproved for the intended build, upload, publication, delivery, or archival workflow

Human checkpoints

Checkpoint	Human responsibility	System responsibility
Intake authority	Define the source, objective, Bloom level, asset, boundaries, preservation rules, and required deliverables.	Validate routing, task mode, package scope, dependencies, and source sufficiency before authoring.
Human checkpoint 1	Review the first governed candidate and approve or reject entry into Nursing Module Audit.	Route feedback to the owning authoring skill and rerun affected local gates before resubmission.
Lead checkpoint	Review the exact Nursing Module Audit-verified version and approve or reject entry into Whole-Module Nursing QA.	Send Lead findings through the applicable text, video, or assessment batch and repeat Nursing Module Audit.
CEO checkpoint	Perform the final human review of the Whole-Module Nursing QA-verified release candidate and accept it or request another governed revision.	Classify CEO's feedback, reopen all affected authoring, Nursing Module Audit, batch, Lead, Whole-Module Nursing QA, regression, and release gates, and return a new exact candidate.
Blocker authority	Resolve missing evidence, conflicting instructions, protected-content decisions, or required scope changes at any stage.	Stop only the affected path, preserve verified work, and resume from the interrupted gate after resolution.

What “output” means at each stage

Output level	Who sees it	Status
Working draft	Specialist and local quality loop	Not final. Still being compared, scored, repaired, or validated.
Audit output	Orchestrator, batch skill, and human when useful	Evaluation evidence and remediation instructions; not the finished learner-facing artifact.
Verified asset	Batch and whole-module ecosystem	Passed its local gate but may still fail cross-asset progression, redundancy, terminology, or distribution review.
Authoring candidate	Initial human reviewer	Passed authoring gates and is eligible to enter Nursing Module Audit.
Nursing Module Audit-verified module	Lead	Exact module version reached Nursing Module Audit 100 and is eligible for Lead review.
Lead-approved module	Whole-Module Nursing QA	Approved by the Lead to enter the independent eight-domain final-QA ecosystem.
Whole-Module Nursing QA release candidate	CEO	All applicable Whole-Module Nursing QA domains reached 100 and release validation passed.
CEO-accepted final	Downstream audience or production workflow	Approved by the CEO for the intended use. Later changes create a new version and governance cycle.

Human and governed-system swimlanes

Each approval moves the exact version into the next QA layer.

Human lane

DefineSources, objectives, constraints, deliverables

Initial reviewApprove for Nursing Module Audit

Lead reviewApprove for independent Whole-Module Nursing QA

CEO reviewAccept or request governed revision

Use finalBuild, upload, publish, deliver, or archive

Governed system

AuthorPlan, create, compare, revise, validate

Nursing Module AuditScore, batch-remediate, rebuild, re-audit to 100

Prepare Whole-Module Nursing QAFreeze Lead-approved exact version

Whole-Module Nursing QAEight domains, remediation, regression, validation

ReleaseDeliver exact CEO-review candidate or iterate

Final-output rule: Authoring approval permits Nursing Module Audit. Nursing Module Audit 100 permits Lead review. Lead approval permits independent Whole-Module Nursing QA. Whole-Module Nursing QA 100 permits CEO review. Only the CEO's acceptance creates the final output.

Four rules for reopening checks after a change: always return to the owning skill and repeat local checks; repeat Nursing Module Audit when a module asset changes; repeat affected Whole Module Audit domains and regression when the whole-module candidate changes; and obtain new human approval when a human-approved version changes.

How a person actually operates the ecosystem

Some actions occur automatically inside a skill. Separate workflows and file handoffs are deliberately started by the human.

Choose the entry point

What you need	What the human runs	What the system does next
Create a text page	Correct L2, L3, L4, or L5 Text Page Writer	Authors, compares, evaluates, repairs, and rechecks the page; embedded interactions and visuals enter their own supporting loops.
Create a video page	Correct level-specific Video Script Author	Authors and audits the script/page, then routes approved narration through segmentation, scene descriptions, graphics, and production QA.
Create an embedded H5P or NGN	Correct L2, L3, L4, or L5 H5P/NGN Designer	Builds and validates the interaction in the context of its text or video page, then enters module-level assessment review.
Create a standalone NGN bank	NGN Question Bank Author	Maps the complete bank to evidence, then routes it through evidence-first audit, blind re-solving, reconciliation, and packaging.
Create a cumulative MCQ bank	Correct Nursing MCQ Level 2, 3, 4, or 5 skill	Authors the bank and routes it through the separate evidence-first MCQ audit. It does not enter Nursing Module Audit.
Run primary module audit	Nursing Module Audit	Audits the complete current module, creates the Nursing Module Audit workbook pair, and routes findings through the three remediation batches until the exact version reaches 100.
Run independent final module QA	Whole-Module Nursing QA Orchestrator	Runs only after Lead approval, executes eight independent domains, challenges scores, coordinates remediation, and validates the CEO-review release candidate.
Repair module findings	QA Batch for Text Pages, Batch QA Video, and QA Batch for Assessments	Each batch reads the workbooks, routes every finding to its owning specialist, verifies the repair, and rebuilds the content master.
Produce instructional visuals	Relevant description designer followed by the approved image-generation skill	Descriptions are evaluated before generation; generated outputs enter a separate visual QA and targeted regeneration loop.

What each connector means

Automatic call	A governing skill invokes a required internal check or subskill without another human prompt.
Human trigger	The person deliberately starts a separate workflow, normally after reviewing an output or receiving a workbook.
File handoff	State moves through a content master, workbook, script, description document, bank, package, or revision record.
Quality gate	A pass or fail decision determines whether work advances or returns to revision.
Blocker	The system stops the affected path because evidence, authority, a dependency, or a required decision is missing.

Entry and exit for each workflow

Workflow	Entry	Exit
Text	Objective, Bloom level, source, module context	Authoring-ready page and worksheet, followed by separate Nursing Module Audit scoring
Video	Objective, source, level, page and production requirements	Authoring-ready video package, followed by separate Nursing Module Audit scoring
Embedded H5P/NGN	Page objective and taught content	Valid interaction in local page and module context
Standalone NGN	Objectives, evidence, Bloom, package specification	Evidence-reconciled, blind-verified bank package
MCQ	Objectives, source, Bloom, item and set requirements	Adversarially verified cumulative bank
Nursing Module Audit	Complete current content master and governed assets	Nursing Module Audit-verified module at 100 for Lead review, or an explicit blocker report
Whole-Module Nursing QA	Lead-approved Nursing Module Audit-verified exact version	Eight-domain-verified release candidate for CEO review

The literal module workflow

This is the operational path between separate runs.

HUMAN STARTSUploads the current content master, objectives, evidence, constraints, and preserve or do-not-touch rules

↓ human trigger

Nursing Module AuditFreezes the source version, audits every governed asset, scores the module, and creates the Nursing Module Audit evaluation and remediation workbooks

↓ file handoff

Evaluation workbook + remediation workbookThe durable control record: asset, evidence, failed criterion, required repair, owner, route, score, and release status

↓ human starts the appropriate batch workflows

QA Batch for Text PagesText, embedded interaction, worksheet, terminology, and page-visual findings

Batch QA VideoScript, page, interaction, segmentation, description, and graphics findings

QA Batch for AssessmentsEmbedded H5P and NGN validity, Bloom, key, feedback, and distribution findings

↓ automatic specialist routing inside each batch

Rebuilt content masterOnly required repairs are incorporated; images, formatting, approved content, filenames, and unrelated assets remain unchanged unless a finding requires a change

Nursing Module Audit OPERATING LOOP: Trigger: any open Nursing Module Audit finding or changed governed asset. Return to: the correct batch and specialist. Repeat: repair, rebuild, and Nursing Module Audit until the exact version reaches 100. Exit: Lead approval permits Whole-Module Nursing QA; only a later Whole-Module Nursing QA-verified candidate reaches CEO.

Files carry the system state

Artifact	Created by	Consumed by	Purpose
Content master	Human and production team	Whole-module and batch workflows	Authoritative collection of learner-facing module assets.
Nursing Module Audit workbook pair	Nursing Module Audit	Three batch skills, human reviewer, and Lead	Records formal Nursing Module Audit scores, exact findings, repair routes, and finding closure.
Whole-Module Nursing QA workbook pair	Whole-Module Nursing QA	Whole-Module Nursing QA remediation routes and CEO	Records eight-domain evidence, challenge-verified scores, cross-domain findings, remediation, and release validation.
Revised asset	Owning specialist	Local audit and batch verifier	Contains the smallest valid repair and preserves accepted work.
Rebuilt master	Batch or Whole-Module Nursing QA remediation workflows	Nursing Module Audit or Whole-Module Nursing QA	Reassembles verified repairs for the appropriate regression and re-audit loop.
Release package	Whole-Module Nursing QA	CEO	Contains the Whole-Module Nursing QA-verified exact candidate, Nursing Module Audit and Whole-Module Nursing QA evidence, remediation status, manifest, limitations, and blocker status.

Source-of-truth hierarchy

1 · Current explicit human direction

2 · Approved objective and authoritative evidence

3 · Governing skill contract

4 · Nursing Assessment Language Standard

5 · Accepted audit findings

6 · Existing module conventions

7 · AI inference

Conflict rule: A lower source cannot silently override a higher source. A material conflict becomes a blocker or human decision point.

Quality-gate hierarchy

Passing one gate never implies that every later gate has passed.

ValidatorFiles, versions, dependencies, contracts

Local assetObjective, Bloom, evidence, language

InteractionCues, key, responses, feedback

ProductionSegmentation, descriptions, visuals, package

BatchFindings resolved and master rebuilt

Whole moduleProgression, consistency, distribution

Human acceptanceFit for intended real-world use

Which gates reopen after a change

Change	Required rechecks	Why
Grammar or formatting only	Language, formatting, preservation	Confirm the edit is clean and did not alter clinical meaning or protected layout.
Clinical claim or nursing action	Evidence, safety, scope, timing, escalation, reassessment, cross-asset consistency	A narrow clinical change may affect several learner-facing assets and assessments.
Learning objective or Bloom level	Authoring, Bloom, page, interaction, assessment, distribution, whole module	The governing target changed, so every dependent asset must be remapped.
H5P, NGN, or MCQ key	Evidence, response validity, distractors, rationale, feedback, blind re-solve, set consistency	Changing the key changes the logic of the entire item, not only one cell.
Video narration	Script audit, page alignment, interaction mapping, segmentation, scene descriptions, affected graphics	Production assets must remain synchronized with the approved narration.
Image, icon, overlay, or label	Description match, accuracy, accessibility, text purpose, caption safety, continuity	The visual may change what the learner sees, infers, or can access.
Terminology rule	Every affected text, video, H5P, NGN, and MCQ asset	Shared language governance prevents corrected terms from being reintroduced elsewhere.

What the system can and cannot guarantee

Can systematically enforce	Still requires human authority
Objective and Bloom alignment	Final instructional intent and stakeholder preference
Evidence mapping and clinical consistency	Local institutional policy when it is not supplied
Language, scope, terminology, and prohibited wording	Approval to change protected or politically sensitive content
Redundancy, progression, assessment validity, and distribution	Whether the release candidate fits the real delivery environment
Traceable findings, repair ownership, and recheck status	Final acceptance, publication, or downstream use

Human feedback that routes cleanly

Weak feedback	Governable feedback
“Make it better.”	“Shorten the L4 explanation without removing the cue-comparison reasoning.”
“Fix the images.”	“Preserve all existing images except Scene 4; regenerate Scene 4 because the icon does not reinforce the teaching text.”
“The questions feel wrong.”	“Recheck Items 8–12 for multiple defensible answers and repeated abstract L4 stem wording.”
“Change the video.”	“Revise the narration in Scene 6, then reopen segmentation, description, and graphics gates only for affected scenes.”

What happens when the system is blocked

A blocker is controlled behavior, not a failed attempt to guess.

Required authority, evidence, objective, file, dependency, or version is missing or conflicting

↓

Stop only the affected release pathPreserve completed work and do not invent the missing information

↓

Issue an exact blocker recordWhat is missing · why it matters · which outputs are withheld · what completed work remains valid · what human decision or file resolves it

↓ human resolves the blocker

BLOCKER RECOVERY LOOP: Trigger: required evidence, authority, file, decision, or tool state is unavailable. Return to: the interrupted gate after the human resolves the blocker. Repeat: affected dependency checks only. Exit: the gate has sufficient proof to continue; unaffected verified work remains closed.

Worked complete-module example

One module can open several local loops, two module-level audit loops, and three human checkpoints.

Human submits Module 8 content masterObjectives, text pages, video pages, embedded interactions, visuals, sources, and preservation rules

↓

Nursing Module Audit finds three connected problemsL4 text page reteaches L2 content · video uses inconsistent terminology · embedded NGN has an ambiguous keyed response

↓ findings split by owner

Text loopL4 writer removes reteaching, preserves analysis, compares with L2 page, and repeats Nursing Module Audit to 100

Video loopVideo writer corrects terminology; narration change reopens affected segment, description, and graphics checks

Assessment loopNGN designer repairs the response set; validity audit blind re-solves and reconciles the item

↓ verified repairs are reassembled

Rebuilt Module 8 master returns to Nursing Module AuditAsset scores, progression, redundancy, terminology, assessment validity, visuals, contradictions, preservation, and scoring integrity run again until Nursing Module Audit reaches 100

↓

Lead reviews the Nursing Module Audit-verified moduleThe Lead approves the exact version for independent Whole-Module Nursing QA

↓

Whole-Module Nursing QA runs eight independent domainsSource-blind score challenge identifies a remaining cross-asset progression weakness; remediation repairs it and replays affected domains until all reach 100

↓ Whole-Module Nursing QA release validation passes

CEO receives Release Candidate 1Completed master, Nursing Module Audit and Whole-Module Nursing QA evidence, remediation status, exact-version record, limitations, and package contents

↓ the CEO requests a shorter L4 explanation but protects the images and NGN

CEO ITERATION LOOP: Trigger: the CEO requests a shorter L4 explanation while protecting other assets. Return to: the L4 text owner. Repeat: only affected objective, Bloom, language, Nursing Module Audit, Lead, Whole-Module Nursing QA progression, regression, and release gates. Exit: the new exact candidate returns to the CEO; verified NGN, video, and protected images remain unchanged.

Operating-model rule: This is a governed network of specialist workflows connected by explicit human triggers, automatic internal calls, durable file handoffs, named quality gates, and revision loops. It is not one continuously running application and it does not silently decide that its own output is final.

Who does what

These are different responsibilities, not different names for “the AI.”

Role	Responsibility	Boundary
Human requester	Defines the exact objective or page goal, Bloom level, asset type, audience, constraints, preserve/do-not-touch rules, accepted decisions, and available source or context required by the selected workflow.	Approves the first governed candidate for Nursing Module Audit but does not create final release authority.
Lead reviewer	Reviews the exact Nursing Module Audit-verified module after the Nursing Module Audit remediation loop reaches 100.	Approves entry into independent Whole-Module Nursing QA or returns findings through the Nursing Module Audit batch loop.
CEO	Performs final human review after all applicable Whole-Module Nursing QA domains reach 100 and release validation passes.	Accepts the exact final output or reopens every governance layer affected by the requested change.
Orchestrator	Determines which audits and routes must run, reconciles results, manages dependencies, and issues workbooks and release status.	Does not replace the specialist skills that author or repair individual assets.
Specialist author or designer	Creates or repairs one defined asset type at one governed cognitive level.	Does not make the final whole-module release decision.
Audit skill	Compares the actual asset with the governing rubric and records criterion-specific evidence and exact findings.	Does not silently rewrite the asset or award a perfect score from template language.
Batch skill	Reads audit workbooks, identifies the owning skill, applies targeted remediation routes, verifies repairs, and rebuilds the master.	Does not replace the level-specific writer or designer.
Validator	Checks required files, contracts, dependencies, versions, package structure, deterministic rules, and release preconditions.	Does not independently establish clinical or instructional quality.
Whole-module auditor	Judges progression, redundancy, consistency, clinical language, assessment validity, distribution, and scoring integrity across the full module.	Does not substitute for the local asset-level audit loop.

How an asset changes governance state

Each state authorizes entry into the next layer; only the last state is final.

Draft

↓

Authoring-ready candidate

↓ initial human approval

Nursing Module Audit

Nursing Module Audit STATE LOOP: Trigger: any score below 100, open finding, changed asset, or workbook defect. Return to: the matching batch and exact owning skill. Repeat: repair, local verification, rebuild, dependency regression, and Nursing Module Audit. Exit: the exact version reaches 100 with no open work.

Nursing Module Audit-verified module

↓ Lead approval

Independent Whole-Module Nursing QA

Whole-Module Nursing QA STATE LOOP: Trigger: any failed domain criterion, signal, regression, workbook, or release check. Return to: the exact owner plus required local and Nursing Module Audit requalification. Repeat: dependency-ordered repair and affected-domain replay. Exit: every applicable domain reaches 100 and release validation passes.

Whole-Module Nursing QA-verified release candidate

↓ CEO acceptance

CEO-accepted final output

Plain-language glossary

Skill	A governed executable work contract combining a defined job, task modes, authority hierarchy, references, exemplars, templates, structured records, builders, validators, quality gates, blocker behavior, exact outputs, and downstream handoff rules—not merely a prompt.
Audit	Evidence-based comparison of the actual asset with fixed criteria.
Batch	A coordinator that remediates many identified assets by routing each finding to its proper owner.
Targeted repair	Preserve correct work and modify only the failed criterion and its necessary dependencies.
100	No changes pending, with asset-specific evidence supporting every perfect criterion.
Evidence-first	Claims and keys must be independently supported rather than accepted because the original writer intended them.
Synchronized release	Connected skills, validators, standards, evidence, workbooks, and packaged outputs agree before release.
H5P	An interactive learning activity embedded in a page or video, such as a decision, classification, sequence, or response task.
NGN	A Next Generation NCLEX-style interaction or case that requires learners to use clinical judgment with evidence from a scenario.
MCQ	A multiple-choice question. In this system, cumulative MCQ banks have their own authoring, evidence, adversarial-review, and release workflow.
SENSE	The governed instructional structure used by applicable video-script skills to organize learner noticing, interpretation, and reasoning.
NURSE	The nursing video reasoning framework: Notice, Understand, Recognize, Select, and Evaluate.
Bloom level	The required cognitive job: L2 Understand, L3 Apply, L4 Analyze, or L5 Evaluate. The level controls what learners must think and do.
Source fingerprint	A stable identifier for the exact collection of files being checked. A changed file creates a changed governed version.
Regression check	A repeat test of the repaired area and its dependencies to prove the fix did not create a new defect elsewhere.
Evidence-ineligible	A claim, placeholder, description, stale file, or unverified output that cannot be used to prove a quality criterion passed.
Nursing Module Audit	Nursing Module Audit: the first formal module-level scoring and batch-remediation ecosystem.
Whole-Module Nursing QA	Whole-Module Nursing QA: the independent final-QA ecosystem that runs after Lead approval and before CEO review.
CJMM	Clinical Judgment Measurement Model. A standalone unfolding case follows the separate CJMM ecosystem; a heading-only CJMM Case Study label is recorded but is not itself a scored asset.

Side-by-side comparison of Nursing Module Audit as primary production assurance and Whole Module Audit as independent system assurance, separated by Lead review as a human gate.

Nursing Module Audit is its own formal QA ecosystem

It begins after the first human approves the authoring candidate and ends before the separate Lead checkpoint.

BEFORE Nursing Module Audit · HUMAN CHECKPOINT 1

Initial human approves the governed authoring candidateThe exact complete module may enter Nursing Module Audit; this approval is not a module score and is not final release.

↓ human-approved exact version

INSIDE THE NURSING MODULE AUDIT ECOSYSTEM

Nursing Module AuditFreezes the exact source version · inventories substantive text and video pages and embedded H5Ps/NGNs · checks objectives, Bloom, clinical accuracy, language, visuals, progression, redundancy, contradictions, production readiness, and criterion-specific evidence

↓ synchronized file handoff

Nursing Module Audit evaluation workbook + Nursing Module Audit remediation workbookFormal scores · asset-specific evidence · exact findings · repair owner · content to preserve · acceptance test · release status

↓ Nursing Module Audit findings

QA Batch for Text PagesRoutes findings to text writers and supporting page skills

Batch QA VideoRoutes findings to script, interaction, segmentation, scene, and graphics skills

QA Batch for AssessmentsRoutes embedded H5P/NGN findings to the correct designer

Nursing Module Audit REMEDIATION LOOP: Trigger: any criterion below 100, open finding, changed asset, or workbook defect. Return to: the matching batch and exact owning specialist. Repeat: atomic repair, actual-asset verification, complete-master rebuild, and Nursing Module Audit on the new exact version. Exit: every applicable criterion equals 100, every finding is closed, both workbooks validate, and no change is pending.

↓ Nursing Module Audit score 100

Nursing Module Audit-verified moduleExact current content master · synchronized assets · validated Nursing Module Audit workbook pair · closed findings · version and preservation record

↓ Nursing Module Audit ecosystem ends

AFTER Nursing Module Audit · HUMAN CHECKPOINT 2

Lead review is a separate human decisionIf the Lead requests a change, the work returns through the applicable Nursing Module Audit batch and Nursing Module Audit re-audit loop. If the Lead approves, the exact Nursing Module Audit-verified version becomes eligible to enter Whole-Module Nursing QA.

Nursing Module Audit and batch skill descriptions

Skill	What it does	How it interacts
Nursing Module Audit	Evaluates individual text pages, video pages, and their embedded H5P/NGN interactions against the asset-level rubric.	Acts as the inner audit gate for text and video authoring. Its findings return to the same level-specific writer until the asset passes.
Nursing Assessment Language Standard	Provides the shared learner-facing rules for nursing terminology, stems, options, prompts, feedback, scope, and prohibited wording.	Governs text, video, H5P, NGN, and MCQ skills so a term corrected in one ecosystem is not reintroduced by another.
QA Batch for Text Pages	Reads text-page findings, identifies the failed criterion and Bloom level, and routes each repair to the correct L2–L5 writer or page-asset skill.	Preserves correct content, applies the smallest valid repair, calls Nursing Module Audit again, and repeats until the page reaches 100.
Batch QA Video	Reads video-page findings and routes them to the correct script writer, page assessment designer, segmenter, scene writer, or graphics skill.	Rechecks the repaired video page through Nursing Module Audit and returns unresolved visual or interaction findings to the exact owning skill.
QA Batch for Assessments	Reads H5P and embedded NGN findings and routes them to the matching L2–L5 H5P/NGN designer.	Repeats validity, Bloom, evidence, answer-key, and feedback checks until the interaction clears its local page context and returns to Nursing Module Audit.

Boundary rule: Nursing Module Audit does not contain the Lead review, and it does not run Whole-Module Nursing QA. The Lead is a human checkpoint after Nursing Module Audit and before Whole-Module Nursing QA. The CEO is not part of this stage.

Whole-Module Audit: independent system assurance

This is the assurance layer above Nursing Module Audit and the standalone NGN and MCQ evidence audits. It evaluates how the verified parts function together as one complete learner experience.

Higher assurance layer. Whole-Module Audit begins only after the applicable module-level and assessment-level assurance work is complete and the Lead approves the exact version. Its 12 subskills independently reopen the source, challenge cross-asset coherence, route affected repairs back through the correct lower assurance tier, and produce a candidate for separate CEO review.

BEFORE Whole-Module Nursing QA · Nursing Module Audit ECOSYSTEM COMPLETE

Nursing Module Audit-verified exact moduleNursing Module Audit score 100 · closed Nursing Module Audit findings · validated Nursing Module Audit workbook pair · synchronized module version

↓

HUMAN CHECKPOINT 2 · LEAD REVIEW

The Lead independently reviews the moduleThe Lead is not a Whole-Module Nursing QA subskill and does not create the Whole-Module Nursing QA score.

Does the Lead authorize Whole-Module Nursing QA?

LEAD LOOP: Trigger: the Lead returns an exact finding. Return to: the applicable text, video, or assessment batch and owner. Repeat: repair, rebuild, and Nursing Module Audit to 100. Exit: the Lead approves the new exact version for Whole-Module Nursing QA.

YES

Lead-approved Whole-Module Nursing QA inputThe exact Nursing Module Audit-verified version is frozen as Whole-Module Nursing QA's starting source.

↓ Whole-Module Nursing QA begins here

INSIDE Whole-Module Nursing QA · 12 AUDIT AND CONTROL SUBSKILLS

Whole-Module Nursing QA OrchestratorCreates an independent source index and manifest, runs the 12 subskills below, validates every internal handoff, reconciles findings and scores, and owns the separate Whole-Module Nursing QA score ledger.

Whole-Module Nursing QA preflight before any subskill can score

Preflight control	What Whole-Module Nursing QA requires	Why it matters
Complete current source	Content master plus every referenced learner-facing text page, video page or script, exact H5P and eligible NGN, visual evidence, captions, alt text, objectives, and authoritative learner order.	Nursing Module Audit workbooks may be evidence, but they never substitute for reopening and independently checking the actual source assets.
Deterministic source index	Every nonblank paragraph, table record, and media file receives a stable record ID and source hash.	Every score, quote, visual claim, finding, and remediation task can be traced to the exact source version.
Normalized manifest	Module identity, source-set fingerprint, exact objective text, authoritative sequence, stable asset/interaction/visual/concept IDs, expected coverage, and one disposition for every indexed record.	Prevents missing rows, duplicate coverage, stale evidence, and audit records that silently disappear.
Scored boundary	Only substantive text pages, substantive video pages or scripts, their embedded H5Ps or eligible NGNs, and their visuals are scored.	Heading-only labels, dividers, navigation, empty shell sections, Walk With Me, and CJMM Case Study are recorded but do not create scores or deductions.
Sequence authority	Use an explicit source-declared order when present; otherwise use complete Content Master document order.	A separate sequence file is optional and is never a blocker when the Content Master already establishes learner order.

How the 12 subskills actually run

Preflight and freeze the exact Whole-Module Nursing QA sourceBuild source index and manifest · establish scored scope and sequence · validate expected coverage · bind one source fingerprint

↓ the orchestrator starts all eight audit domains internally

1 · Asset learningText and video teaching

2 · H5P/NGN validityEach interaction

3 · Option validityL3–L5 responses

4 · Assessment setWhole-set distribution

5 · Visual productionActual visual evidence

6 · Clinical languageSafety, scope, terminology

7 · ProgressionPrior learning and growth

8 · RedundancyRecurring teaching jobs

↓ eight validated domain handoffs

9 · Source-blind semantic score challengeHide proposed scores · reopen exact source · independently rejudge every criterion · withhold anything not proved

↓ challenge-verified evidence only

Cross-domain synthesisMerge root causes · prevent duplicate deductions · reconcile contradictions · build change-impact and dependency order

↓ stable consolidated finding ledger

10 · Remediation plannerOne atomic task per finding · exact owner · content to preserve · acceptance test · verification · regression scope

↓ approved execution route

11 · QA executor and specialist repairChange the owning asset · verify actual revised evidence · preserve unaffected accepted work

↻ REQUIRED REQUALIFICATION WHEN AN ASSET CHANGES
Trigger: Whole-Module Nursing QA repair changes a governed asset → Return to: owning specialist → Repeat: affected local gates and Nursing Module Audit on the changed exact version when applicable → Exit: the requalified exact asset returns to Whole-Module Nursing QA

↓ repaired and requalified exact version

Whole-Module Nursing QA regression replayRe-run the owning domain plus dependent assessments, contradictions, progression, redundancy, distribution, and score

↓ no open governed work

12 · Workbook compilation and release validationBuild both workbook deliverables · reconcile every ledger and handoff · run pre-release validation · reopen files · run final read-only validation

↻ Whole-Module Nursing QA CLOSED LOOP
Trigger: failed criterion, unresolved signal, missing proof, workbook defect, or release check → Return to: exact earlier owner → Repeat: repair, requalification, dependent-domain regression, workbook reconciliation, and release validation → Exit: Ready or an explicit Not ready / Score withheld state

Automatic-routing rule: The human starts Whole-Module Nursing QA once. The orchestrator creates or loads the internal domain handoffs, validates and repairs them, runs the semantic challenge, synthesizes findings, and builds the workbooks. The user is never asked to manually invoke twelve subskills or supply internal handoffs.

The 12 Whole-Module Nursing QA subskills and what each one does

Subskill	What it checks	Evidence and anti-inflation controls	Successful output	Failure route
1 · Whole-Module Asset Learning Audit	Every substantive text and video asset's objective, intended Bloom, learner task, completeness, instructional function, clinical teaching, and relationship to its parent module.	Requires one source-bound evidence row per expected asset, every governed criterion exactly once, affirmative passing proof, boundary testing, and a unique evidence key.	Validated Asset Content & Learning handoff with exact coverage, source-bound criterion evidence, proposed scores, findings, and signals ready for semantic challenge.	Returns atomic findings to the owning L2–L5 text or video author and reopens dependent page, assessment, progression, redundancy, and Whole-Module Nursing QA checks.
2 · Whole-Module H5P/NGN Validity Audit	Every exact embedded H5P and eligible NGN: objective mapping, Bloom, scenario necessity, cues, responses, key, feedback reasoning, and parent-page fit.	Requires one row for every manifest interaction ID. A parent-page or bank-summary row cannot stand in for its child interactions.	Validated Individual H5P/NGN Validity handoff containing exactly one complete evidence row for every manifest interaction ID.	Routes through QA Batch for Assessments to the exact L2–L5 H5P/NGN designer and reopens the containing page when context changes.
3 · Whole-Module Option Validity Audit	L3–L5 response options for multiple defensible answers, weak or implausible distractors, cueing, asymmetry, shortcut answering, cognitive-level collapse, and keyed reasoning.	Independently tests the easiest legitimate answer path, strongest competitor, counterevidence, and whether one answer remains clearly best.	Validated L3–L5 Option Validity handoff with one source-bound option judgment, strongest-competitor analysis, findings, and signals per applicable interaction.	Returns the affected interaction or assessment to its level-specific writer and rechecks key, rationale, feedback, item validity, and related set-level effects.
4 · Whole-Module H5P/NGN Set Audit	The complete embedded assessment set for objective coverage, Bloom and learning-job distribution, repeated scenarios, cue saturation, answer-pattern concentration, feedback overload, assessment-order leakage, and unused taught sibling topics.	Uses one scored MODULE-SET row plus a complete source-bound semantic fingerprint inventory for every interaction; inventory items do not receive duplicate set scores.	Validated H5P/NGN Set Distribution handoff with one MODULE-SET score row and a reconciled fingerprint inventory for every interaction.	Routes only the items needed to repair the set through QA Batch for Assessments, then reruns individual and set-level validity.
5 · Whole-Module Visual Production Audit	Whether required page and video visuals are present, accurate, readable, accessible, instructionally purposeful, synchronized, caption-safe, clinically safe, and production-ready.	Uses indexed media or structural hashes and actual rendered evidence; a description, placeholder, or filename cannot substitute for the final visual file.	Validated Visual/Production handoff covering every expected visual with actual-file evidence, accessibility and production judgments, findings, and signals.	Routes to the description designer, page-image generator, segmenter, scene writer, graphics skill, or manual production repair according to the root cause.
6 · Whole-Module Clinical Language Audit	Clinical accuracy, evidence support, safety, entry-level RN scope, terminology, timing, escalation, reassessment, policy dependence, and contradictions across text, video, interactions, and visuals.	Requires exact source references, tests competing interpretations and policy dependence, and separates a shared root cause from its cross-modal consequences.	Validated Nursing Scope/Clinical Safety/Language handoff with source-specific evidence, cross-modal contradiction records, findings, and clinical-verification routes.	Routes to every affected asset owner, the shared language standard, or manual clinical verification; dependent domains replay after repair.
7 · Whole-Module Progression Audit	Whether prior learning advances across the authoritative sequence—from foundations through application, analysis, and evaluation—without gaps, invalid entry assumptions, or higher levels restarting from scratch.	Every progression row requires earlier and later asset IDs, increasing sequence positions, exact references for both, learner tasks, reasoning change, and a controlled decision.	Validated Progression/Coverage handoff with exact earlier-later comparisons, learner entry states, reasoning changes, decisions, findings, and regression scope.	Returns to the earliest upstream asset causing the progression gap and every dependent page or assessment affected by the change.
8 · Whole-Module Cross-Asset Redundancy Audit	Every recurring concept family across text, video, examples, H5Ps, and NGNs; distinguishes purposeful reinforcement from repeated full beginner teaching.	Uses one complete module-set concept ledger, canonical and later source references, teaching-job comparison, restart-from-scratch test, allowed-repetition reason, and compression decision.	Validated Cross-Asset Redundancy/Streamlining handoff with one complete concept ledger, one comparison for every recurring family, and exact compression decisions.	Routes the later duplicate to compress as a bridge, remove, or rewrite for progression while preserving the canonical teaching owner.
9 · Source-Blind Semantic Score Challenge	Rejudges every governed criterion after hiding all proposed numeric scores, ratings, and prior audit conclusions.	Reopens the exact indexed source; records a 0–4 verified score, affirmative proof, boundary test, counterevidence, rationale, unique evidence key, and source refs. Only this challenge score can enter Whole-Module Nursing QA.	Challenge-verified criterion scores and discrepancy decisions—or an explicit withheld score when the exact source cannot prove the criterion.	Withholds an unproved score, rejects boilerplate or repeated evidence, reconciles discrepancies, and returns invalid handoffs for repair before synthesis.
10 · Whole-Module Remediation Planner	The consolidated finding ledger after all eight domains validate; groups root causes, detects conflicts, orders dependencies, and creates one executable repair task per finding.	Every task requires exact location, one required change, content to preserve, route, dependencies, acceptance criteria, verification method, regression scope, and matching source fingerprint.	Validated dependency-ordered remediation roadmap with one atomic, traceable, executable task for every finding.	Routes each atomic task to the owning specialist. A proposed repair never closes a finding and the planner never adds points.
11 · Whole-Module Nursing QA Executor	Implements an approved remediation roadmap or verifies revised assets when the user asks for execution.	Requires actual revised evidence—not suggested wording—and checks that every repair satisfies its acceptance test without damaging protected content.	Verified revised assets, completion evidence, updated finding status, and the exact regression scope ready for Nursing Module Audit requalification and Whole-Module Nursing QA replay.	Replays the affected domain, dependent assessments, contradiction checks, progression, redundancy, and score; unresolved work returns to the owning route.
12 · Workbook Compilation and Release Validation	Builds the separate Whole-Module Nursing QA evaluation and remediation workbook pair, reconciles all handoffs and ledgers, and runs the final pre-release plus read-only validation.	Checks sheet order and schemas, source fingerprint, exact coverage, score recomputation, weights, primary score ownership, finding-remediation reconciliation, atomicity, execution routes, regression, consumer cells, data-only readability, formula errors, signals, and readiness.	Validated Whole-Module Nursing QA evaluation and remediation workbooks plus Ready, Not ready, or Score withheld status bound to the exact source version.	Any failed check blocks release, leaves the score blank or status Not ready/Score withheld as governed, and returns the specific defect to the proper earlier subskill.

How the twelve are counted: Eight are independent audit-domain skills. Four are mandatory control subskills: source-blind semantic score challenge, remediation planning, remediation execution/verification, and workbook compilation with release validation. The orchestrator coordinates them but does not replace their separate checks.

Whole-Module Nursing QA scoring weights and what a score of 100 means

Domain	Weight	100 requires
Asset Content & Learning	15%	Every applicable criterion verified at 4/4 with source-specific proof and no open asset-learning work.
Individual H5P/NGN Validity	15%	Every exact interaction independently covered and fully valid.
L3–L5 Option Validity	10%	Every applicable response set supports one defensible best answer without shortcuts or ambiguity.
H5P/NGN Set Distribution	10%	The complete module set is balanced, distinct, and free of open distribution findings.
Visual/Production	10%	All required actual visuals and production evidence pass.
Nursing Scope/Clinical Safety/Language	15%	No open clinical, scope, safety, terminology, timing, escalation, reassessment, or contradiction finding.
Progression/Coverage	15%	The authoritative sequence demonstrates supported cognitive progression without gaps.
Cross-Asset Redundancy/Streamlining	10%	Every recurring concept has a justified teaching job or an executed streamlining repair.

Ready means all conditions agree: Complete-module scope; all required handoffs valid; every applicable domain and the validated Whole-Module Nursing QA score equal 100; no open finding, unresolved signal, fingerprint mismatch, remediation conflict, pending verification, or failed regression; and both workbooks pass final validation. Missing evidence is left blank and produces Score withheld—it is never converted to zero or estimated.

Whole-Module Nursing QA evaluation workbook

Executive Summary	Validated score, readiness, blockers, plain-English eight-domain scorecard, and actual required changes.
Governed Score Ledger	Weights, coverage, recomputed domain scores, findings, signals, and validation.
Eight domain detail tabs	Asset, interaction, option, set, visual, clinical, progression, and redundancy evidence.
Assessment QA Handoff	Diagnostic bridge to QA Batch for Assessments; never claims authoritative Nursing Module Audit item scores.
Finding Ledger + Remediation Roadmap	Atomic findings, owners, dependencies, preservation, acceptance, verification, and regression.
Rubric and Score Caps + Release Validation	Visible scoring rules and check-specific release evidence.

Whole-Module Nursing QA remediation workbook

README	Plain operating instructions and status meaning.
Content To Do	Text and video work routed to the owning skills.
H5P To Do	Embedded interaction work routed to assessment QA.
Finding Remediation	One-to-one source of truth matching the evaluation Finding Ledger.
Visual To Do	Filtered visual and production repairs.
Clinical Verification	Every open clinical-language finding requiring source or human verification.
Verification Log	Actual completion evidence, regression results, and finding closure.

Run all supported domain handoffs and source-blind challengesValidate exact coverage · reject stale fingerprints · reconcile primary score ownership · consolidate root causes

↓

All applicable domains at 100 and every release check passed?

Plan and execute atomic repairsRoute to owning skills · preserve unaffected content · verify actual revised evidence

Whole-Module Nursing QA LOOP: Trigger: the readiness gate fails. Return to: the exact owner and required local/Nursing Module Audit requalification. Repeat: affected domain, dependent assessments, contradictions, progression, redundancy, score, workbook reconciliation, and release validation. Exit: Ready or a correctly documented blocker/withheld-score state.

YES

Whole-Module Nursing QA-verified release candidateExact module · separate Whole-Module Nursing QA workbook pair · completed remediation evidence · manifest · version record · release-validation result

↓ Whole-Module Nursing QA ecosystem ends

AFTER Whole-Module Nursing QA · HUMAN CHECKPOINT 3

The CEO performs a separate final human reviewThe CEO is not a Whole-Module Nursing QA subskill and does not create the Whole-Module Nursing QA score. The CEO accepts the exact candidate for intended use or requests a change that reopens every affected governance gate.

Lead review, Whole-Module Nursing QA, and CEO review are three separate authorities

Stage	Type	Input	Decision or output	What it cannot claim
Lead review	Human checkpoint before Whole-Module Nursing QA	Nursing Module Audit-verified exact module at 100	Approves entry into Whole-Module Nursing QA or returns the module through the Nursing Module Audit remediation loop.	Does not run Whole-Module Nursing QA, create its score, or create final acceptance.
Whole-Module Nursing QA	Independent governed QA ecosystem	Lead-approved Nursing Module Audit-verified exact version	Runs 12 subskills and produces a Whole-Module Nursing QA-verified release candidate only after every governed release condition passes.	Does not replace Lead judgment or CEO's final human decision.
CEO review	Human checkpoint after Whole-Module Nursing QA	Whole-Module Nursing QA-verified exact release candidate	Accepts the exact version as final for intended use or requests a governed revision.	Does not retroactively become part of Whole-Module Nursing QA; a requested change creates a new version and required rechecks.

Separation rule: Nursing Module Audit 100 permits Lead review. Lead approval permits Whole-Module Nursing QA. Whole-Module Nursing QA Ready permits CEO review. Only the CEO's acceptance creates the final output.

What happens when a text-page skill runs

The writer owns authoring readiness. Nursing Module Audit separately owns the formal score and finding closure.

Responsibility and routing at a glance

Human starts	Runs automatically inside the workflow	Human decision	Output
Choose the L2–L5 writer and provide the exact objective, source, page context, constraints, and protected content.	The writer builds its worksheet, drafts, compares, revises, validates, renders, and packages the exact page; Nursing Module Audit findings route through QA Batch for Text Pages.	Approve the authoring-ready page for Nursing Module Audit; later the Lead and CEO independently accept or return exact versions.	A CEO-accepted page inside the released exact module—not merely an AI draft or writer-ready page.

User runs the correct L2, L3, L4, or L5 Text Page WriterThe assigned objective and Bloom level determine which writer owns the page

↓

Lock the inputsExact MLO/EO or page goal · Bloom level · source content · module context · audit findings · preserve/do-not-touch instructions

↓

Complete the thinking-proof worksheetSource inventory · traceability · conceptual story · Bloom reasoning · clinical and language QA · comparison plan · preservation requirements

↓

Validate, build, render, and inspect the worksheetStructural validator · task-specific DOCX · visual page inspection · exact-version synchronization

↓

Draft at the assigned cognitive levelL2 explains · L3 applies · L4 analyzes · L5 evaluates

↓

Compare, revise, and resynchronizeSource evidence · objective · Bloom contract · text-page language standard · clinical precision · earlier and later pages · embedded H5P/NGN · worksheet · visuals · audit findings

WRITER LOOP: Trigger: the draft, worksheet, validator, render, source comparison, objective, Bloom, clinical-language, or page-context check fails. Return to: the same L2–L5 writer and the exact failed upstream component. Repeat: targeted repair, worksheet synchronization, semantic checks, validation, rebuild, render, and inspection. Exit: the writer's readiness gate passes; the writer does not assign the Nursing Module Audit score.

↓

Authoring-ready page packageExact current page candidate · synchronized worksheet · validator and render evidence · fingerprint · readiness verdict · unresolved findings

↓ initial human approval for Nursing Module Audit

Nursing Module Audit formally evaluates the exact page in the moduleClinical accuracy · scope · objective alignment · Bloom fidelity · completeness · progression · redundancy · terminology · assessment alignment · visual readiness

↓

Nursing Module Audit score 100 with every finding closed?

NO · a criterion failed

QA Batch for Text Pages routes the exact findingsSame level-specific writer, H5P/NGN designer, worksheet component, terminology authority, or page-visual skill

↓

Make the smallest valid repairPreserve correct content; change only what failed; check for regression, new redundancy, scope drift, and contradictions

TEXT REMEDIATION LOOP: Trigger: Nursing Module Audit records a text, embedded-interaction, terminology, worksheet, or page-visual finding. Return to: QA Batch for Text Pages and the exact owning specialist. Repeat: smallest valid repair, local readiness checks, master rebuild, dependency regression, and Nursing Module Audit on the new exact version. Exit: the page reaches 100 and every applicable finding is closed.

YES · all criteria passed

Place the Nursing Module Audit-verified page in the rebuilt modulePage, aligned interaction, worksheet, terminology, visual state, audit evidence, and exact-version identifiers are reconciled

↓

Continue through Lead review, independent Whole-Module Nursing QA, and CEO reviewThe page can still reopen for cross-asset progression, redundancy, terminology, assessment-distribution, production, Lead, Whole-Module Nursing QA, or CEO findings

What each pass does

Loop station	Detailed action	Quality standard enforced
1 · Lock inputs	Reads the exact MLO/EO or page goal, assigned Bloom level, source content, module context, required format, existing audit findings, and any preserve/do-not-touch instructions.	Prevents scope drift, invented objectives, unsupported teaching, accidental image changes, and work outside the requested asset.
2 · Draft	Builds the page using the level-specific teaching job: L2 explains foundations; L3 applies findings in focused nursing situations; L4 analyzes relationships and priorities; L5 evaluates whether care is safe, complete, and effective.	Maintains Bloom fidelity, entry-level RN scope, clinically accurate sequencing, and an appropriate amount of teaching for the page goal.
3 · Compare	Compares every major claim and instructional move with the supplied source, objective, existing module content, Nursing Assessment Language Standard, related page interactions, and any Whole-Module findings.	Detects contradictions, unsupported additions, terminology drift, duplication, missing prerequisite links, and higher-level pages that reteach lower-level content from scratch.
4 · Evaluate	Runs Nursing Module Audit criterion by criterion and records asset-specific evidence for clinical accuracy, objective alignment, Bloom, instructional completeness, learner language, assessment alignment, visual readiness, and release status.	A 100 cannot be assigned from template language or assumption; every perfect criterion requires evidence from the actual page and no pending correction.
5 · Targeted repair	Routes each failed criterion back to the same writer or the exact supporting skill. The repair changes only what failed, preserves correct material, and updates connected H5P/NGN, worksheet, terminology, or visual description only when the finding requires it.	Prevents wholesale rewrites, regression, unnecessary page growth, broken formatting, and new inconsistencies introduced while fixing a narrow problem.
6 · Re-audit	Re-runs the affected criteria and the dependency checks, confirms the repair did not damage previously correct content, reconciles all findings, and repeats the full loop if any issue remains.	Release occurs only at 100 with no changes pending; the verified page then returns to the batch skill and ultimately to Whole-Module Nursing QA for cross-asset review.

Text-page skill descriptions

Skill	What it owns	Comparison and revision role
Nursing L2 Text Page Writer	Creates foundational pages that help learners explain, distinguish, classify, organize, sequence, or interpret essential nursing concepts.	Compares the page with the exact objective and source, checks foundational completeness without application-level overreach, repairs failed Nursing Module Audit criteria, and repeats until 100.
Nursing L3 Text Page Writer	Creates application-focused pages in which learners use taught cues, rules, and nursing actions in focused clinical situations.	Checks that the scenario is necessary, that learners must apply rather than recall, and that the page does not reteach all L2 content; revises and re-audits until 100.
Nursing L4 Text Page Writer	Creates analysis-focused pages that connect cue clusters, competing explanations, priorities, safety, and QSEN-style rescue reasoning.	Compares relationships across findings instead of listing facts, removes abstract wording and lower-level reteaching, repairs exact Nursing Module Audit failures, and re-runs the audit until 100.
Nursing L5 Text Page Writer	Creates evaluation-focused pages in which learners judge whether a plan or response is safest, complete, effective, timely, and appropriately reassessed.	Checks the explicit criteria used for judgment, evidence sufficiency, escalation and reassessment logic, then revises failed criteria and re-audits until 100.
Nursing Level 2 H5P/NGN Designer	Builds page interactions that interpret, classify, sequence, or distinguish taught L2 content.	Compares the interaction with the page’s exact objective and taught evidence; revises cues, options, key, and feedback until local and whole-module validity gates pass.
Nursing Level 3 H5P/NGN Builder	Builds focused application interactions that require learners to use page content in a necessary scenario.	Checks scenario necessity, action selection, option validity, and feedback reasoning; returns failures to the interaction and page context for another audit.
Nursing Level 4 H5P/NGN Designer	Builds analysis interactions involving cue relationships, pattern changes, prioritization, or competing clinical interpretations.	Checks that the analysis cannot be answered by recall, that all cues are functional, and that keyed reasoning is evidence-grounded; revises until the validity loop passes.
Nursing Level 5 H5P/NGN Designer	Builds evaluation interactions requiring judgment against explicit safety, completeness, effectiveness, timing, or reassessment criteria.	Compares every response with the stated criteria and evidence, repairs ambiguity or unsupported judgment, and repeats validity review until release.
Instructional Image Description Designer	Creates production-ready descriptions for standalone instructional page images from the page objective and taught content.	Checks that each visual teaches a necessary concept, is clinically safe and feasible, matches the page, and has no unsupported anatomy, procedure, product, or text requirement.
Instructional Text Page Images	Generates the final standalone instructional image only from a complete approved description.	Returns any inaccurate, unreadable, noncompliant, or instruction-mismatched image to regeneration; the final page cannot receive visual completion credit until the actual image is present.
QA Batch for Text Pages	Coordinates remediation across every text page identified in the whole-module workbooks.	Routes each finding to the exact writer or page-asset skill, protects correct content and images, verifies the revised page with Nursing Module Audit, and repeats until 100.
Nursing Module Audit	Provides the governing asset-level rubric and evidence requirements used inside every text-page loop.	Compares the actual page with the objective, Bloom level, clinical and language standards, interaction alignment, and visual state; it releases only when no criterion still needs repair.

Concrete Level 4 text-page cycle

The Nursing L4 Text Page Writer locks the L4 objective, source content, existing module progression, and any Nursing Module Audit or Whole-Module findings.
It drafts an analysis page built around relationships among cues, changing patterns, competing priorities, and safe nursing decisions instead of restating foundational facts.
It compares the draft with the source, L4 Bloom requirements, language rules, earlier pages, related H5P/NGN content, and the Nursing Module Audit rubric.
Nursing Module Audit evaluates the actual draft with criterion-specific evidence. Any score below 100 must include a concrete repair instruction.
The L4 writer receives only the failed criteria, makes the smallest valid repair, preserves correct content, and checks that the repair did not create new redundancy, scope drift, or contradictions.
Nursing Module Audit runs again. This closed cycle repeats until the page scores 100 with no changes pending; QA Batch for Text Pages then returns the verified page to the rebuilt module for Whole-Module Nursing QA.

Text-page release gate: 100 means clinically accurate, objective-aligned, Bloom-faithful, complete, concise, terminology-compliant, assessment-aligned, visually accounted for, supported by asset-specific evidence, and free of unresolved findings.

Video-page authoring and quality loop

Script, page, interaction, segmentation, description, and graphics work remain separate but connected quality loops.

Responsibility and routing at a glance

Human starts	Runs automatically inside the workflow	Human decision	Output
Choose the correct standard or LJL video writer and provide the objective, Bloom level, sources, framework, assessment requirements, and filename.	The writer compares and repairs the script/page; embedded interactions route to their designers; Nursing Module Audit findings route through Batch QA Video.	Approve the exact script handoff to production; later the Lead and the CEO separately review the governed video package.	An approved exact script handoff first; after production and governance, a CEO-accepted learner-facing video page and package.

Run the correct level-specific video script skillLock objective · Bloom · sources · SENSE requirements · assessment mapping · exact filename

↓

Draft script and complete learner-facing video pageIntro · transition · closing · narration · SENSE · embedded H5P/NGN

↓ compare against source, objective, language standard, module progression, and page context

Pre-production script and page audit + Batch QA VideoEvaluate script, page, clinical accuracy, Bloom, redundancy, interaction alignment, SENSE or NURSE requirements, and readiness for the approved-script production handoff. This does not replace the final audit of the actual produced package.

↓

All script and page criteria passed?

Route the exact findingScript writer · blurb writer · H5P/NGN designer · segmenter · scene writer · graphics skill

Targeted repairPreserve correct components and repair only the owning part

INNER VIDEO LOOP: Trigger: any script, page, interaction, SENSE/NURSE, clinical, Bloom, redundancy, or supporting-asset criterion fails. Return to: the exact owning video or interaction skill. Repeat: targeted repair, source and page comparison, dependent checks, Batch QA Video, and Nursing Module Audit on the new exact version. Exit: every script/page criterion passes and the approved exact script handoff is issued.

YES

APPROVED EXACT SCRIPT HANDOFFVideo authoring ends; Video production beginsThe approved narration, filename, objective, Bloom level, SENSE or NURSE requirements, and assessment boundaries are locked for segmentation and scene production

Audit the actual produced package before releaseAfter segmentation, descriptions, and images are complete, Nursing Module Audit and Batch QA Video evaluate the exact rendered video package to 100 → Lead review → independent Whole-Module Nursing QA → CEO final review.

Video ecosystem skill descriptions

Skill	What it does	How it compares and loops
Nursing Level 2 Video Script Author	Authors L2 scripts that help learners explain, distinguish, organize, sequence, or interpret nursing content using the governed SENSE structure.	Compares narration and SENSE elements with the exact objective, source, learner level, and language rules; repairs failed criteria and rechecks until the script and page pass.
Video Level 3 Script	Authors application-level scripts from fixed outcomes, source content, design notes, and assessment requirements.	Checks that learners use taught knowledge in a focused situation, that the scenario is necessary, and that embedded assessment aligns; revises and re-audits until release.
Nursing Level 4 Video Script Author	Authors analysis-level scripts and embedded interactions involving cue relationships, changing patterns, prioritization, and rescue reasoning.	Compares each segment with L4 evidence and scenario logic, removes lower-level reteaching, repairs invalid cues or reasoning, and repeats Nursing Module Audit.
Nursing Level 5 Video Script Author	Authors evaluation-level scripts requiring judgment of safety, completeness, effectiveness, timing, escalation, and reassessment.	Checks that judgments use explicit criteria and sufficient evidence; repairs unsupported conclusions and re-runs the quality gate.
LJL Nursing Level 3 NURSE Video Script Author	Creates L3 scripts using the project NURSE sequence: Notice, Understand, Recognize, Select, and Evaluate.	Checks that each NURSE phase performs a distinct application job, aligns to the objective, and does not duplicate teaching; failed phases return to targeted revision.
LJL Nursing Level 4 NURSE Video Script Author	Creates L4 analysis scripts using NURSE to connect cues, interpretations, choices, and evaluation.	Compares each phase with analysis-level evidence and removes steps that collapse into recall; revisions repeat until phase and whole-script audits pass.
LJL Nursing Level 5 NURSE Video Script Author	Creates L5 evaluation scripts using NURSE to judge whether nursing decisions meet explicit clinical criteria.	Checks evidence sufficiency, decision criteria, escalation, and reassessment; returns unsupported judgments to the writer until the audit clears.
Nursing Video Blurb Writer	Creates the learner-facing introduction, transition, and closing while preserving the exact script filename.	Compares the blurb with the video’s actual objective and level, removes promises the video does not fulfill, and returns mismatches to revision.
Video Script Segmenter	Removes headings and assessment blocks, preserves the complete approved narration, and divides it into production-ready stationary-image scenes.	Checks source-to-segment completeness, visual-state stability, sentence integrity, and excluded material; resegments until the approved production contract passes.
Visual Transition Script Segmenter	Segments narration into visually distinct scene blocks for downstream storyboard and image-description work.	Compares adjacent scenes for a real visual change, prevents long static holds or arbitrary splits, and repeats segmentation when a scene has no distinct visual job.
Caption-Safe Scene Writer	Turns approved scene text into production-ready descriptions with layout, cast, teaching text, icon decisions, and caption-safe composition.	Classifies each scene as teaching overlay or illustration-only, tests whether text teaches, evaluates icon candidates for conceptual reinforcement, and revises descriptions that fail.
QA Caption-Safe Video Scenes	Orchestrates segmentation, scene-description QA, text-purpose checks, icon decisions, and release scoring.	Loops rejected scenes back to the segmenter or scene writer until the complete package passes, then authorizes graphics production.
Caption-Safe Illustrated Graphics	Generates caption-safe static video artwork from approved descriptions and character references.	Uses batch manifests, inspects every image, protects accepted files, and regenerates rejected or missing scenes only.
Batch QA Video	Routes whole-module video findings to the exact script, page, interaction, segmentation, description, or graphics skill.	Re-runs Nursing Module Audit after targeted repair and continues routing until the video page and package have no unresolved findings.
L2–L5 H5P/NGN Designers	Create Bloom-matched interactions embedded in video pages.	Compare cues, responses, key, and feedback with the script and objective; invalid interactions return through QA Batch for Assessments and the video-page audit.
Nursing Module Audit	Evaluates the complete learner-facing video page, not only the narration.	Checks script, page, objective, Bloom, clinical language, interaction alignment, progression, and supporting asset readiness; findings return to the owning video skill.

Video-page release gate: The learner-facing page and script clear Nursing Module Audit; embedded interactions pass their own validity loop; segmented narration is complete; descriptions are approved; and every generated scene passes batch QA.

Video production is a separate ecosystem after script approval

It converts one approved exact script into a reconciled segmented script, approved descriptions, separate generated scene files, and a verified production package. It is not the video-authoring loop and it is not the standalone text-page-image loop.

Responsibility and routing at a glance

Human starts	Runs automatically inside the workflow	Human decision	Output
Hand off the approved exact script and choose the governing segmenter contract.	QA Caption-Safe Video Scenes coordinates segmentation, reconciliation, description QA, scoring, packaging, and targeted returns; graphics production generates and inspects scene batches.	Authorize the production run and resolve escalations; the Lead and CEO review only after the actual package re-enters module governance.	A reconciled segmented script, approved scene descriptions, accepted individual images, manifest, and governed video-production package.

APPROVED EXACT SCRIPT HANDOFF0 · Video production beginsLock narration · filename · objective · Bloom · SENSE or NURSE framework · excluded headings · excluded H5P blocks · source fingerprint

↓ authority selection before segmentation

Which one segmenter contract governs this production run?

VIDEO SCRIPT SEGMENTER

Remove headings and complete H5P blocksPreserve every retained narration word exactly once and in order; never split a sentence; isolate SENSE when the selected contract requires it

VISUAL TRANSITION SEGMENTER

Preserve all source wording, including headingsUse for a route that intentionally retains headings; do not treat it as interchangeable with the remove-headings production contract

Segmentation standard: The production route removes excluded headings and H5P blocks, isolates SENSE when required, preserves every retained word, and applies the stationary-image visual-state and clinical-presentation gates.

↓ segmented-script candidate

1 · Reconcile segmentation against the exact sourceEvery retained word exactly once · original order · headings and assessments handled by the selected contract · complete sentences · consecutive numbering · one stable image state per scene

↻ SEGMENTATION LOOP
Trigger: missing, duplicated, reordered, broken, or visually unstable scene text → Return to: selected segmenter → Repeat: split or merge failed boundaries, score, and source-to-segment reconciliation → Exit: at least 9.5/10 with no fidelity or visual-state defect

↓ approved scene text blocks

2 · Caption-Safe Scene WriterCreate one independent production description per scene with stable cast, setting, action, clinical presentation, exact visible text, layout family, text-purpose decision, three-candidate icon test, title treatment, and lower caption-safe area

↓ description set

3 · Description and package QA layerReconcile every scene with the segmented script · classify teaching-overlay versus illustration-only · reject purposeless text · require NO ICON when no candidate teaches · verify title, SENSE, cast, anatomy, one-frame stability, references, filenames, and package completeness

↻ DESCRIPTION LOOP
Trigger: a scene, text-purpose, icon, title, cast, anatomy, caption-safe, filename, or package check fails → Return to: Caption-Safe Scene Writer or segmenter → Repeat: revise rejected descriptions, rescore, and reconcile the full set → Exit: the complete description package reaches its release threshold

Automatic production routing: QA Caption-Safe Video Scenes coordinates segmentation QA, description QA, icon and text decisions, scoring, reconciliation, and documents-only packaging without requiring the user to invoke each internal check separately.

↓ approved descriptions and one cast identity reference

4 · Caption-Safe Illustrated GraphicsCreate the title image, exactly one approved cast thumbnail showing the complete cast facing the viewer, and every scene through separate image-generation calls; manage requests in batches; save every image as a separate file; persist accepted files and manifest state immediately

↓ each generated file is inspected before the next batch advances

5 · Per-image production QADescription match · exact boxed text · purposeful icons · illustration-only fidelity · cast and room continuity · clinically congruent expression · coherent anatomy and hands · upper overlay zone · lower caption clearance · no collage · expected filename

↻ BATCH IMAGE LOOP
Trigger: a generated file is rejected, missing, corrupt, mismatched, or newly fails regression → Return to: graphics generation for that file only → Repeat: regenerate, inspect at full size, preserve accepted finals, and reconcile manifest and count → Exit: every required scene is accepted and the manifest is complete

Image-production standard: Use the batch-capable caption-safe contract: reserve the lower 28–32%, isolate identity references, inspect clinical presentation, protect accepted files, and resume only rejected, missing, or newly failed scenes.

↓ verified final files only

6 · Assemble the production packageApproved segmented script · approved image-description document · exactly one approved cast thumbnail containing the complete video cast facing the viewer · title image · every accepted scene image · final manifest · rejection or regeneration history when required

↓ return the actual package, not descriptions alone

7 · Video governance re-entryBatch QA Video routes production findings when that route is available → Nursing Module Audit evaluates the actual rendered video page and production evidence → Lead → Whole-Module Nursing QA visual and production regression → CEO

↻ GOVERNANCE RETURN LOOP
Trigger: any material script, description, image, package, Nursing Module Audit, Lead, Whole-Module Nursing QA, or CEO finding → Return to: the earliest owning step affected by the change → Repeat: all dependent production checks plus applicable Nursing Module Audit and Whole-Module Nursing QA regression → Exit: the new exact package again clears every reopened gate

Video-production skills and responsibilities

Skill or layer	Owns	Does not own
Video Script Segmenter	Remove headings and H5Ps, preserve retained narration, create and score stationary-image scenes, and enforce one stable visual state.	Does not write descriptions or generate images.
Visual Transition Script Segmenter	Preserve all supplied source wording and create visually distinct scene text blocks when headings are intentionally retained.	Does not replace the production route that must remove headings and assessments.
Caption-Safe Scene Writer	Write production-ready descriptions and apply text, icon, layout, cast, caption, clinical-presentation, and one-frame gates.	Does not generate final images unless downstream generation is explicitly included.
QA Caption-Safe Video Scenes	Orchestrate segmentation QA, scene-description QA, scoring, reconciliation, and documents-only packaging.	Does not generate the final artwork.
Caption-Safe Illustrated Graphics	Generate and inspect separate caption-safe images in resumable batches while protecting accepted files.	Does not decide script fidelity or repair narration.
Batch QA Video	Route Nursing Module Audit video findings to the exact script, assessment, segmentation, description, or graphics owner.	Does not replace any specialist owner.
Nursing Module Audit	Score the actual video page, production evidence, visuals, embedded interactions, clinical language, progression, and readiness.	Does not replace upstream authoring or production skills.
Whole-Module Nursing QA visual-production lane	Independently challenge whether actual visuals are present, accurate, accessible, purposeful, synchronized, and production-ready.	Does not author scenes or generate images.

Video-production release gate: The authoritative segmenter contract is unambiguous; retained narration reconciles exactly; descriptions pass purpose, icon, clinical, continuity, and caption-safe checks; every separate generated file passes full-size inspection; the package manifest is complete; and the actual video page returns through Nursing Module Audit and Whole-Module Nursing QA after any material change.

Standalone text-page image ecosystem

This tab owns page-image descriptions, explicit generation authorization, one-file generation, inspection, insertion, and page re-audit. Caption-safe video scenes are now broken out in the separate Video production tab.

Responsibility and routing at a glance

Human starts	Runs automatically inside the workflow	Human decision	Output
Provide the approved page, objective, placement, evidence boundaries, and visual teaching need.	The description designer compares, scores, and revises the specification; after explicit approval, the image skill generates and inspects one file at a time.	Explicitly approve the description for generation, then approve the actual inserted image through page and module review.	One accepted instructional image placed in the page with accessibility evidence, followed by Nursing Module Audit on the rendered page.

Standalone instructional page imageBegins with an approved text page, objective, placement, and a production-ready image description. Produces one instructional image file that returns to the text page and Nursing Module Audit.

Separate video-production boundaryBegins only after an approved video script and uses segmentation, description, image-generation, batch-QA, package, and governance-return loops. The complete chain is in the Video production tab.

Text-page image flow

1 · Lock the teaching jobApproved page content · exact objective · Bloom level · image placement · learner need · source boundaries · accessibility need

↓

2 · Instructional Image Description DesignerSelects the correct visual type and specifies the scene, composition, labels, exact text, accessibility, restrictions, clinical boundaries, and what the visual must teach

↓ compare with page, evidence, image rules, and learning purpose

3 · Description evaluation and purposeful revisionChecks objective alignment, necessity, feasibility, cognitive load, medical safety, accessibility, exact wording, placement, and independence from assessment answers

↓

Description scores at least 90 and is explicitly approved?

Return exact description findingsRevise purpose, content, layout, labels, restrictions, or accessibility instructions

DESCRIPTION LOOP: Trigger: purpose, objective, evidence, feasibility, load, clinical safety, accessibility, wording, or placement fails. Return to: Instructional Image Description Designer. Repeat: revise, recompare with the page and evidence, and rescore. Exit: at least 90 plus explicit APPROVED FOR IMAGE GENERATION status.

YES

4 · Human invokes Instructional Text Page ImagesOne approved description authorizes one separate image; approval is not inferred from a draft description

↓

5 · Generate and inspect the actual fileOpen at original size; inspect clinical accuracy, legibility, composition, exact text, accessibility, instruction match, and file integrity

↓

Does the actual image pass?

Regenerate only this imagePreserve the approved description unless the failure proves the description itself is defective; use the bounded retry path

IMAGE LOOP: Trigger: the actual file fails instruction match, clinical accuracy, text, composition, accessibility, or integrity. Return to: image generation—or the description designer if the specification caused the defect. Repeat: record, regenerate only the failed file, and reinspect. Exit: the actual image passes; otherwise escalate at the governed retry limit.

YES

6 · Insert into the text pageConfirm placement, alt text, rendering, surrounding explanation, and that the page still teaches without contradiction or answer leakage

↓

7 · Nursing Module Audit + QA Batch for Text PagesThe rendered page—not the description alone—must pass visual integration and every connected criterion at 100; a failed image reopens the containing page and affected module checks

Image ecosystem skill descriptions

Skill	What it owns	How it compares and loops
Instructional Image Description Designer	Designs and scores the complete specification for one standalone instructional page image before generation is allowed.	Compares the visual with the exact page objective, source evidence, learning purpose, placement, accessibility, cognitive load, clinical restrictions, and feasibility; requires purposeful revision and explicit generation approval.
Instructional Text Page Images	Generates one original instructional image from one complete, approved description.	Inspects the actual file at full size for accuracy, readability, exact instruction match, accessibility, and technical integrity; regenerates only the failed image and escalates after the bounded retry path.
Nursing Module Audit	Evaluates visual integration in the actual learner-facing text page.	A description or generated file cannot earn release by itself; the rendered page must support the objective, avoid contradiction and answer leakage, meet accessibility and placement requirements, and clear the applicable criteria at 100.

Standalone page-image release gate: The description reaches at least 90 and receives explicit generation authorization; the separately generated file passes full-size inspection; the image is inserted with accurate accessibility and placement evidence; and the actual rendered text page clears Nursing Module Audit at 100.

NGN evidence, validity, and independent-review assurance

This is a formal assurance ecosystem at the same governance tier as Nursing Module Audit. It independently tests NGN evidence, clinical and cognitive validity, cueing, scoring, item function, set quality, and release readiness.

Peer assurance tier—not an asset-production tab. NGN authoring creates the questions. This ecosystem challenges the exact authored version through evidence verification, cueing analysis, blind independent re-solving, reviewer reconciliation, set-level checks, and package validation.

What the NGN evidence audit contains

Assurance layer	Specific checks	Required evidence or output
Objective and source evidence	Maps the scenario, exhibits, cues, task, response model, key, rationale, and feedback to the exact objective, Bloom level, learner scope, and approved evidence.	Traceable item-level evidence map with no unsupported clinical claim, response, or scoring decision.
Clinical and cognitive validity	Checks clinical accuracy, safety, entry-level RN scope, scenario necessity, cue sufficiency, reasoning demand, and whether the item truly performs its stated L2–L5 cognitive job.	One defensible reasoning path at the intended Bloom level without requiring knowledge that was not taught or sourced.
Cueing and answer leakage	Tests whether wording, option patterns, repeated terminology, unequal specificity, exhibit structure, grammatical fit, response counts, or feedback language reveals the key or allows test-wise guessing.	A cue-free learner version in which the correct response is supported by clinical evidence rather than surface clues.
Response model and scoring	Verifies the NGN format, selectable elements, response limits, keyed combinations, partial-credit logic when applicable, rationale consistency, and feedback for correct and incorrect pathways.	Synchronized learner item, faculty key, rationale, feedback, and scoring specification.
Blind independent review	A reviewer re-solves the item without seeing the original key or author rationale, records the selected response and reasoning, and identifies ambiguity, missing cues, competing keys, or unsupported assumptions.	Blind-solve record kept separate from the original-key review until comparison.
Reviewer reconciliation	Compares the authored key, evidence audit, and blind solution; investigates every disagreement instead of resolving it by majority vote or deference to the author.	Documented resolution, targeted repair when needed, and a new blind check on the repaired exact version.
Item-function and measurement quality	Reviews cognitive load, likely difficulty, discrimination risk, distractor or response functioning, construct-irrelevant complexity, fairness, accessibility, and format fit before administration.	Pre-administration item-function judgment. Statistical psychometrics are added only when learner-response data are available.
Set and package assurance	Checks objective coverage, format and cognitive distribution, repeated assessment jobs, redundancy, answer-pattern risk, workbook synchronization, file integrity, and exact-version identifiers.	Validated bank or embedded-item package with every finding closed and every component synchronized.

Responsibility and routing at a glance

Human starts	Runs automatically inside the workflow	Human decision	Output
Select standalone-bank or embedded-page scope and provide objective, Bloom level, evidence, learner scope, and package requirements.	Authoring maps every element to evidence; the evidence-first audit verifies, blind re-solves, reconciles, and routes atomic repairs to the owning NGN skill.	Release a validated standalone bank, or approve an embedded NGN to continue through assessment batch, Nursing Module Audit, Lead, Whole-Module Nursing QA, and CEO.	A synchronized evidence-qualified bank, or a governed embedded interaction inside the final module.

Run NGN Question Bank Author or the correct L2–L5 H5P/NGN DesignerLock objective · Bloom · evidence · learner scope · standalone or embedded package route

↓

Author scenario, exhibits, cues, tasks, responses, keys, rationales, and feedbackEvery learner-facing element is mapped to evidence and the exact objective

↓

Evidence-First NGN Question Bank AuditVerify evidence → test clinical and cognitive validity → screen cueing and answer leakage → validate response model and scoring → review item function → blind re-solve without the original key → reconcile disagreements

↓

Evidence, key, Bloom, clinical safety, cue sufficiency, cueing, response model, scoring, feedback, blind review, reconciliation, item-function, set, and package gates passed?

Return item-level findings to the owning NGN skillExact evidence gap, ambiguous response, invalid cue, Bloom mismatch, or blind-review disagreement

Remediate and remapRepair only the failed item components and preserve verified evidence

INNER NGN LOOP: Trigger: evidence, key, Bloom, cue, response, feedback, blind-solve, reconciliation, or package gate fails. Return to: the exact owning NGN author or level-specific designer. Repeat: atomic repair, evidence remap, blind re-solve, reconciliation, and affected set/package validation. Exit: every item and workbook/package gate passes.

YES

Release standalone bank or continue the embedded routeStandalone banks exit through their evidence-first human-review route. Embedded NGNs continue through QA Batch for Assessments and Nursing Module Audit.

Embedded governance continuesNursing Module Audit reaches 100 → Lead review → independent Whole-Module Nursing QA validity and set review → CEO final review

NGN ecosystem skill descriptions

Skill	What it does	How it compares and loops
NGN Question Bank Author	Builds original standalone or unfolding-case NGN banks from exact objectives, Bloom levels, learner scope, and approved evidence.	Maps each cue, exhibit, task, response, key, and rationale to evidence; repairs findings from blind re-solving and evidence audit until the package is synchronized and releasable.
Evidence-First NGN Question Bank Audit	Audits, remediates, blind re-solves, reconciles, and packages NGN banks using evidence-grounded keyed reasoning.	Separates original-key review from blind solving, records disagreements, requires reconciliation, and returns every failed item to the author before final release.
Nursing Level 2 H5P/NGN Designer	Creates L2 interactions requiring interpretation, classification, sequencing, or distinction of taught evidence.	Checks exact page objective and taught content; revises any cue, response, key, or feedback that exceeds or misses L2.
Nursing Level 3 H5P/NGN Builder	Creates L3 application interactions in necessary focused scenarios.	Compares the required nursing action with taught rules and scenario cues; repairs items answerable by recall or unsupported inference.
Nursing Level 4 H5P/NGN Designer	Creates L4 interactions based on cue relationships, pattern changes, or competing interpretations.	Checks cue necessity, relationship analysis, response validity, and keyed reasoning; returns ambiguity or recall-level tasks to redesign.
Nursing Level 5 H5P/NGN Designer	Creates L5 interactions requiring judgment against explicit safety, completeness, effectiveness, timing, or reassessment criteria.	Compares every choice with the stated criteria and evidence; unsupported judgments loop back to authoring.
QA Batch for Assessments	Routes embedded H5P/NGN findings to the exact level-specific designer.	Rechecks local interaction validity, page context, and whole-module set distribution after every repair.
Whole-Module H5P/NGN Validity Audit	Checks every embedded interaction individually and as a set across the module.	Detects objective gaps, repeated assessment jobs, invalid responses, weak feedback, and distribution problems; findings return through the assessment batch loop.
Nursing Assessment Language Standard	Controls learner-facing scenario, prompt, response, rationale, and feedback language.	Prevents terminology drift and prohibited wording from reappearing during remediation.

NGN release gate: Evidence mapping, objective and Bloom fidelity, clinical safety, cue sufficiency, cueing and answer-leakage screens, response-model validity, keyed reasoning, feedback, blind independent re-solve, disagreement reconciliation, item-function review, set distribution, workbook evidence, and package validation must all pass on the same exact version.

MCQ evidence, psychometric quality, and independent-review assurance

This is a formal assurance ecosystem at the same governance tier as Nursing Module Audit. It independently tests evidence, one-best-answer validity, cueing, distractor function, Bloom demand, psychometric quality, bank distribution, and release readiness.

Peer assurance tier—not an asset-production tab. L2–L5 MCQ writers create the items. This ecosystem challenges the exact bank through deterministic screens, cueing analysis, blind and adversarial solving, nurse-educator review, disagreement reconciliation, psychometric or item-function review, and synchronized release validation.

What the MCQ evidence audit contains

Assurance layer	Specific checks	Required evidence or output
Objective, source, and key evidence	Maps the stem, clinical cues, keyed option, distractors, rationale, and nursing decision to the exact objective, Bloom level, learner scope, and approved source.	Evidence-supported key with no unsupported assumption and a documented reason every distractor is inferior.
One-best-answer validity	Tests whether the stem is answerable as written, all necessary information is present, the key is uniquely best, distractors are plausible but clearly inferior, and the rationale matches the actual item.	One defensible best answer under the stated conditions, with no second answer that becomes correct through a reasonable interpretation.
Cueing and test-wiseness	Screens option length, specificity, grammar, repeated words, absolute language, answer-position patterns, implausible distractors, paired options, category mismatch, and rationale leakage that can reveal the key without nursing reasoning.	A cue-free item in which success depends on the intended nursing judgment rather than test-taking shortcuts.
Blind and adversarial review	An independent reviewer solves the item without the original key, then deliberately argues for the strongest competing option and tests the item aloud for ambiguity, hidden assumptions, and language defects.	Separate blind-solve and adversarial-review records completed before the authored key and rationale are revealed.
Reviewer reconciliation	Compares the authored key, evidence screen, blind solution, adversarial challenge, and nurse-educator review; every disagreement is investigated and repaired at its source.	Documented resolution and re-review of the repaired exact item rather than automatic acceptance of the original key.
Pre-administration psychometric review	Evaluates intended difficulty, discrimination risk, distractor functioning, option homogeneity, construct relevance, cognitive load, fairness, accessibility, and alignment between item demand and Bloom target.	Item-function profile and targeted revisions before learner use; this is qualitative psychometric review, not a claim of observed statistical performance.
Post-administration psychometrics	When learner-response data are supplied, examines item difficulty, point-biserial discrimination, distractor selection, reliability contribution, omitted-response patterns, and fairness or differential-functioning signals.	Data-based retain, revise, monitor, or retire decision. This layer is marked not applicable when response data are unavailable.
Bank-level distribution and release	Checks objective and Bloom coverage, key-position balance, topic distribution, repeated scenarios or decision jobs, redundancy, terminology, rationale coherence, document synchronization, and exact-version integrity.	A synchronized cumulative bank with every item and set-level finding closed before human release.

Responsibility and routing at a glance

Human starts	Runs automatically inside the workflow	Human decision	Output
Choose the L2–L5 MCQ writer and provide the objective, Bloom level, source, item count, bank design, and language rules.	The writer authors items; the evidence-first audit runs deterministic screens, adversarial solving, read-aloud review, reconciliation, and targeted returns.	Accept or return the synchronized cumulative bank through its own evidence-release route.	A human-released cumulative MCQ bank; it does not enter Nursing Module Audit or Whole-Module Nursing QA unless a separate contract explicitly changes scope.

Run the correct Nursing MCQ Level 2, 3, 4, or 5 skillLock objective · Bloom · source · item count · set design · language standard

↓

Author stem, four options, one best answer, and rationalesThe level-specific skill controls the cognitive job and nursing-decision demand

↓

Evidence-First Nursing MCQ AuditEvidence and one-best-answer verification → cueing and test-wiseness screens → distractor-function and psychometric review → blind solve → adversarial challenge → nurse-educator read-aloud review → reconciliation

↓

Evidence, one-best-answer, Bloom, cueing, distractor, blind-review, adversarial, psychometric, reconciliation, bank-distribution, and release gates passed?

Return each failure to the matching level-specific MCQ writerKey support · distractor validity · Bloom · wording · rationale · terminology · distribution

Targeted item or set repairChange only the failed element and preserve already verified content

INNER MCQ LOOP: Trigger: key support, distractor, Bloom, wording, rationale, terminology, distribution, adversarial-solve, or reconciliation check fails. Return to: the matching L2–L5 MCQ writer or set owner. Repeat: targeted repair, deterministic screens, adversarial solve, read-aloud review, and bank reconciliation. Exit: every item and set gate passes.

YES

Synchronized cumulative-bank releaseEvidence, keys, options, rationales, Bloom, language, and set distribution are reconciled

Separate ecosystem boundaryCumulative MCQ banks do not route through Nursing Module Audit or Whole-Module Nursing QA

MCQ ecosystem skill descriptions

Skill	What it does	How it compares and loops
Nursing MCQ Level 2	Creates and quality-checks four-option single-best-answer items that assess understanding of nursing concepts.	Compares each item with the exact objective and source, checks that it requires interpretation rather than simple trivia, and revises failures returned by the evidence-first audit.
Nursing MCQ Level 3	Creates application items requiring use of taught nursing knowledge in a focused scenario.	Checks scenario necessity, actionable nursing decisions, cue sufficiency, and plausible distractors; revises items answerable without application.
Nursing MCQ Level 4	Creates analysis items requiring learners to connect cues, explain patterns, or prioritize among competing clinical interpretations.	Removes repeated abstract stem formulas, ensures the analysis drives a nursing decision, and repairs items that collapse into recall or have multiple defensible answers.
Nursing MCQ Level 5	Creates evaluation items requiring judgment of safety, completeness, effectiveness, timing, escalation, or reassessment.	Checks that explicit criteria and adequate evidence support one best answer; ambiguous or unsupported judgment returns to revision.
Evidence-First Nursing MCQ Audit	Audits large keyed nursing banks, remediates defects, independently re-solves items, and verifies set-level quality before release.	Runs deterministic checks and adversarial nurse-educator review, reconciles disagreements, and returns each failure to the matching L2–L5 author until the bank clears.
Nursing Assessment Language Standard	Controls stems, options, rationales, feedback, scope, terminology, and prohibited wording.	Applies the same terminology rules across every item and rechecks repaired items so removed language is not reintroduced elsewhere in the bank.

MCQ release gate: Every item must pass source and key verification, one-best-answer validity, Bloom fidelity, cueing and test-wiseness screens, distractor-function review, blind solving, adversarial challenge, nurse-educator read-aloud review, disagreement reconciliation, pre-administration psychometric review, applicable post-administration statistics, rationale and terminology checks, bank distribution, and exact-version synchronization.

CJMM case-study ecosystem

CJMM case studies use a separate unfolding-case authoring, evidence, validity, remediation, regression, packaging, and human-review loop. A heading-only CJMM label in a module is still recorded but not scored as an asset.

Responsibility and routing at a glance

Human starts	Runs automatically inside the workflow	Human decision	Output
Define the exact objective, clinical-judgment demand, evidence, learner scope, unfolding phases, response formats, and package contract.	Case authoring synchronizes phases; evidence, unfolding-state, validity, and set checks synthesize findings and route atomic remediation plus dependent-phase regression.	Accept the validated case package or request a governed change; module-integrated assets require their own Nursing Module Audit, Lead, Whole-Module Nursing QA, and CEO gates.	A validated exact CJMM package with evidence, scoring, feedback, accessibility, audit, remediation, regression, and version records.

HUMAN STARTS1 · Define the case-study jobExact objective · Bloom or clinical-judgment demand · source evidence · learner scope · unfolding phases · required response formats · package requirements

↓

CJMM AUTHORING LAYER2 · Build the unfolding caseClient state · phase-by-phase cue release · exhibits · prompts · responses · keys or scoring map · rationales · feedback · accessibility

↻ AUTHORING LOOP
Trigger: source, objective, phase, cue, scoring, or feedback defect → Return to: exact case phase or item → Repeat: revise and resynchronize dependent elements → Exit: complete evidence-mapped case candidate

↓ candidate package

3A · Evidence and clinical-source auditVerify every cue, trend, judgment, action, outcome, escalation, key, and feedback statement against taught or approved evidence

3B · Unfolding-state auditVerify chronology, cue timing, hidden-versus-available information, state transitions, and prevention of later-phase leakage

3C · Clinical-judgment validity auditBlind solve · strongest competitor · one-defensible-decision logic · scope · safety · response-format integrity

↓ reconciled audit findings

AUTOMATIC SYNTHESIS4 · Case-set synthesis and remediation planMerge shared root causes · detect repeated cue and decision jobs · order dependencies · create one atomic task per finding · define preservation, acceptance, verification, and regression scope

↓

Did every evidence, phase, item, set, scoring, feedback, accessibility, and package criterion pass?

5A · CJMM remediationRepair the exact failed phase, exhibit, item, response, key, feedback, or package component while preserving verified work

↻ CJMM REMEDIATION LOOP
Trigger: any open finding → Return to: owning case component → Repeat: actual repair, blind re-solve, affected later-phase replay, set regression, and package validation → Exit: no open governed work

YES

5B · Validated CJMM packageExact case · evidence map · scoring map · feedback · accessibility · audit record · remediation evidence · regression result · version record

↓

HUMAN DECISION6 · Human case-study reviewAccept the exact package for its intended use or request a governed change that reopens the affected CJMM loops

↓ accepted exact package

7 · Downstream integrationStandalone case follows its case-study release route. Any governed text, video, H5P, NGN, or visual used inside the module also re-enters its owning asset, Nursing Module Audit, Lead, Whole-Module Nursing QA, and CEO gates when changed.

CJMM functional responsibilities

Functional layer	What it checks or creates	Successful output and next route
CJMM Case Study Authoring	Creates the complete unfolding case, phase states, cue release, exhibits, learner tasks, response formats, scoring map, rationales, feedback, and accessibility.	Returns a synchronized evidence-mapped case candidate to independent case audits.
CJMM Evidence and Clinical-Source Audit	Verifies every clinically meaningful claim, cue, response, decision, outcome, and feedback statement.	Returns source-specific findings or an evidence-qualified case handoff.
CJMM Unfolding-State Audit	Checks chronology, phase boundaries, cue timing, persistent evidence, hidden information, and later-phase leakage.	Returns an exact dependency map showing which later phases and items must replay after a change.
CJMM Clinical-Judgment Validity Audit	Blind solves the case, tests strongest competitors, verifies one-defensible-decision logic, scope, safety, scoring, feedback, and response-format validity.	Returns reconciled validity findings or a challenge-verified case handoff.
CJMM Case-Set Audit	Checks objective coverage, repeated cue families, repeated decision jobs, scoring concentration, item independence, progression, and accessibility across the whole case.	Returns set-level findings without duplicating item-level deductions.
CJMM Remediation Orchestrator	Creates and routes one atomic task per finding with exact location, owner, preservation, dependency, acceptance, verification, and regression requirements.	Returns actual revised evidence for blind re-solving and affected-phase regression.
CJMM Package and Release Validation	Reconciles the exact case, evidence, scores, findings, remediation, accessibility, file structure, and version record.	Returns a human-review candidate only after every required check passes.
Nursing Assessment Language Standard	Checks learner-facing scenario, prompt, response, rationale, feedback, scope, terminology, and prohibited wording without changing the case’s clinical-judgment demand.	Returns language-qualified case elements to the affected validity and regression checks.

CJMM boundary rule: A heading-only CJMM label creates no score. A complete standalone case follows the separate CJMM ecosystem. If a repair changes a module-governed text, video, interaction, or visual, that changed component also returns through its owning local loop and every affected Nursing Module Audit and Whole-Module Nursing QA gate.

Complete people-facing skill and workflow directory

Each skill has one bounded job. The next-route description shows how its output enters the larger ecosystem.

System governance, standards, and remediation

Skill	What it owns	Output and next route
Nursing Module Audit	Primary formal audit of the complete current module: text pages, video pages, embedded H5P/NGN, visuals, clinical language, Bloom, progression, redundancy, production readiness, and scoring evidence.	Returns synchronized evaluation and remediation workbooks. Findings split into the three batch routes; the rebuilt exact version returns to Nursing Module Audit until it reaches 100, then goes to the Lead.
QA Batch for Text Pages	Coordinates remediation for text pages, page worksheets, embedded interactions, terminology, and page visuals without replacing the level-specific writer.	Returns verified repaired pages and a rebuilt master to Nursing Module Audit; preserves images and accepted content unless an explicit finding requires change.
Batch QA Video	Coordinates remediation for video scripts, page blurbs, embedded interactions, SENSE or governed NURSE, segmentation, scene descriptions, graphics, and package completeness.	Routes each finding to its exact owner, verifies the repair, rebuilds the video page/package, and returns the module to Nursing Module Audit.
QA Batch for Assessments	Coordinates remediation for embedded H5P and NGN objective mapping, Bloom, scenario necessity, cue and response validity, keys, feedback, language, and set distribution.	Returns repaired interactions to their page context, Nursing Module Audit, and the applicable whole-module validity and set checks.
Whole-Module Nursing QA Orchestrator	Independent final-QA controller after Lead approval. Freezes a new manifest, runs eight domain lanes, challenges scores, synthesizes root causes, plans remediation, replays regression scope, and validates release.	Returns Whole-Module Nursing QA evaluation and remediation workbooks plus an exact release candidate. Atomic findings go to the owning skills; the candidate goes to the CEO only when all applicable domains reach 100.
Whole-Module Clinical Language Audit	Cross-module clinical accuracy, evidence support, safety, RN scope, terminology, timing, escalation, reassessment, policy dependence, and cross-modal contradictions.	Returns source-specific findings to text, video, assessment, or shared-language owners; affected and dependent Whole-Module Nursing QA domains rerun after repair.
Whole-Module H5P/NGN Validity Audit	Individual validity of every embedded H5P and eligible NGN: objective, Bloom, cues, scenario necessity, responses, key, feedback, clinical safety, and production fitness.	Returns interaction findings through the referenced assessment route and reopens the containing text or video page when context changed. Whole-set distribution is a separate internal Whole-Module Nursing QA lane.
Nursing Assessment Language Standard	Shared learner-facing language authority for MCQ, H5P, and NGN stems, scenarios, prompts, response elements, options, labels, rationales, feedback, and prohibited wording.	Requalifies affected assessments after repair. It does not replace the separate text-page, video, clinical, or whole-module language authorities.

Text-page authoring and embedded interaction skills

Skill	When and what it creates	Comparison, revision, and next route
Nursing L2 Text Page Writer	Creates foundational pages for explaining, distinguishing, classifying, organizing, sequencing, or interpreting essential nursing concepts.	Compares source, objective, L2 Bloom, clinical language, module context, worksheet, interactions, and visuals; revises to authoring readiness, then enters human review and Nursing Module Audit.
Nursing L3 Text Page Writer	Creates application pages in which learners use taught cues, rules, and nursing actions in a focused necessary situation.	Checks scenario necessity and prevents L2 reteaching; exact Nursing Module Audit or Whole-Module Nursing QA findings return to targeted repair and re-audit.
Nursing L4 Text Page Writer	Creates analysis pages connecting cue clusters, changing patterns, competing explanations, priorities, safety, and rescue reasoning.	Checks relationship analysis rather than recall, removes abstract or repetitive framing, preserves accepted content, and reopens affected page and module gates after repair.
Nursing L5 Text Page Writer	Creates evaluation pages requiring judgment of whether care is safe, complete, effective, timely, escalated, and reassessed.	Checks explicit criteria and evidence sufficiency; unsupported judgments return to targeted revision and formal re-audit.
Nursing Level 2 H5P/NGN Designer	Creates L2 embedded interactions requiring interpretation, classification, sequencing, or distinction of taught content.	Checks the exact page objective, taught evidence, responses, key, and feedback; enters page context, QA Batch for Assessments, Nursing Module Audit, and Whole-Module Nursing QA validity routes.
Nursing Level 3 H5P/NGN Builder	Creates L3 embedded interactions requiring application in a necessary focused scenario.	Rejects recall-only or unsupported-inference items; repairs return through local validity, page, batch, Nursing Module Audit, and whole-module checks.
Nursing Level 4 H5P/NGN Designer	Creates L4 embedded interactions requiring cue relationships, changing patterns, prioritization, or competing interpretations.	Checks functional cues and keyed analysis; ambiguity or lower-level demand returns to redesign and every affected validity route.
Nursing Level 5 H5P/NGN Designer	Creates L5 embedded interactions requiring judgment against explicit clinical criteria.	Checks evidence, criteria, response validity, escalation, reassessment, key, and feedback; unsupported judgment returns to redesign and re-audit.

Video authoring and production skills

Skill	When and what it creates	Comparison, revision, and next route
Nursing Level 2 Video Script Author	Creates L2 Understand nursing scripts using the governed SENSE structure.	Compares narration, SENSE, source, objective, learner level, language, and assessment mapping; repairs failures before page audit and production.
Video Level 3 Script	Creates L3 Apply instructional video scripts from fixed outcomes, sources, design notes, and assessment requirements.	Checks necessary application, scenario function, taught actions, and embedded assessment alignment; approved narration proceeds to segmentation.
Nursing Level 4 Video Script Author	Creates L4 Analyze nursing scripts and embedded interactions from fixed objectives, masters, and exemplars.	Checks cue relationships, pattern changes, priorities, and rescue reasoning; narration changes reopen downstream segments, descriptions, and graphics.
Nursing Level 5 Video Script Author	Creates L5 Evaluate scripts using SENSE when learners judge safety, completeness, effectiveness, timing, escalation, and reassessment.	Checks explicit decision criteria and evidence sufficiency; unsupported conclusions return to script revision and every affected downstream gate.
LJL Nursing Level 3 NURSE Video Script Author	Creates L3 video scripts only when the project explicitly requires NURSE: Notice, Understand, Recognize, Select, Evaluate.	Checks that every phase has a distinct application job and maps to the objective; failed phases return to targeted revision.
LJL Nursing Level 4 NURSE Video Script Author	Creates L4 NURSE scripts connecting cues, interpretations, choices, priorities, and evaluation.	Checks analysis-level work in every phase and removes recall-only or duplicative steps before page and production routes.
LJL Nursing Level 5 NURSE Video Script Author	Creates L5 NURSE scripts using evidence and explicit criteria to judge nursing decisions.	Checks sufficiency, safety, escalation, and reassessment; unsupported judgments return to the owning phase and script audit.
Nursing Video Blurb Writer	Creates the learner-facing introduction, transition, and closing around the approved video while preserving its exact filename.	Compares every promise with what the video actually teaches; mismatches return to blurb revision and video-page review.
Video Script Segmenter	Removes excluded headings and assessment blocks and divides the complete retained narration into stationary-image scenes.	Isolates SENSE when required, preserves every retained word, and enforces stable visual-state and clinical-presentation gates.
Visual Transition Script Segmenter	Creates visually distinct scene blocks for storyboard and image-description work.	Rejects arbitrary splits, broken sentences, long static holds, and adjacent scenes without a meaningful visual change.
Caption-Safe Scene Writer	Creates production-ready scene descriptions with composition, cast, teaching text, separate text boxes, icon decisions, and caption clearance.	Classifies teaching overlay versus illustration-only, tests three icon candidates, records NO ICON when needed, and returns failed descriptions to revision.
QA Caption-Safe Video Scenes	Orchestrates segmentation, description scoring, learning-job checks, icon logic, reconciliation, and the documents-only package.	Loops rejected scenes to the segmenter or scene writer until the complete package reaches the governed threshold; then authorizes graphics production.
Caption-Safe Illustrated Graphics	Generates separate caption-safe static scene images from approved descriptions with resumable batch manifests and accepted-file protection.	Keeps the lower 28–32% clear, inspects every image, retains accepted files, and regenerates rejected or missing images only.

Standalone image skills

Skill	What it owns	Output and next route
Instructional Image Description Designer	Designs, evaluates, purposefully revises, and scores the complete production specification for one standalone instructional page image.	Returns a description marked APPROVED FOR IMAGE GENERATION only after objective, evidence, purpose, feasibility, restrictions, accessibility, exact text, and placement pass.
Instructional Text Page Images	Generates one original image from one explicitly approved description and inspects the actual full-size file.	Returns the accepted image to the containing text page; failures regenerate only that file, and the rendered page returns through text-page and Nursing Module Audit visual integration checks.

Standalone NGN and cumulative MCQ skills

Skill	What it owns	Output and next route
NGN Question Bank Author	Builds original standalone or unfolding-case NGN banks from exact objectives, Bloom, learner scope, evidence, and package requirements.	Returns synchronized scenarios, exhibits, cues, tasks, responses, keys, rationales, feedback, and evidence mappings to the evidence-first NGN audit.
Evidence-First NGN Question Bank Audit	Independently verifies evidence, audits validity, blind re-solves without the original key, reconciles disagreements, remediates defects, and validates the package.	Returns a releasable standalone bank or precise item-level findings to the NGN author. Embedded NGNs instead continue through the page, Nursing Module Audit, and Whole-Module Nursing QA routes.
Nursing MCQ Level 2	Creates L2 four-option single-best-answer items requiring nursing understanding rather than trivia.	Returns evidence-mapped items to the evidence-first MCQ audit; failed wording, options, keys, rationales, Bloom, or distribution return to L2 repair.
Nursing MCQ Level 3	Creates L3 application items requiring a necessary focused nursing situation and actionable decision.	Returns items to evidence-first and adversarial audit; recall-only, weak-cue, or invalid-distractor findings return to L3 repair.
Nursing MCQ Level 4	Creates L4 items requiring cue relationships, changing patterns, competing interpretations, or prioritization.	Returns items to evidence-first audit; abstract repetitive stems, recall collapse, and multiple defensible answers return to L4 repair.
Nursing MCQ Level 5	Creates L5 items requiring judgment of safety, completeness, effectiveness, timing, escalation, or reassessment.	Returns items to evidence-first audit; missing criteria, insufficient evidence, ambiguity, or unsupported judgment return to L5 repair.
Evidence-First Nursing MCQ Audit	Runs evidence, language, mechanical, option, and set screens; independently and adversarially re-solves items; performs nurse-educator read-aloud review; reconciles all findings.	Returns precise failures to the matching L2–L5 writer until the cumulative bank is synchronized and releasable. This separate ecosystem does not route through Nursing Module Audit.

CJMM case-study functional layers

Functional layer	What it owns	Output and next route
CJMM Case Study Authoring	Creates the exact unfolding case, client state, phase structure, cue release, exhibits, learner tasks, response formats, scoring map, rationales, feedback, and accessibility.	Returns a synchronized evidence-mapped candidate to the three independent case-audit lanes.
CJMM Evidence and Clinical-Source Audit	Verifies every clinical claim, cue, decision, action, outcome, escalation, key, scoring rule, and feedback statement.	Returns source-specific findings or an evidence-qualified handoff to case synthesis.
CJMM Unfolding-State Audit	Checks chronology, cue timing, hidden information, phase transitions, persistence of earlier evidence, and later-phase leakage.	Returns a dependency map for remediation and affected-phase regression.
CJMM Clinical-Judgment Validity Audit	Blind solves the case and checks strongest competitors, one-defensible-decision logic, scope, safety, formats, scoring, and feedback.	Returns challenge-verified findings or a valid case handoff to case synthesis.
CJMM Case-Set Audit	Checks whole-case coverage, repeated cue and decision jobs, scoring concentration, item independence, progression, and accessibility.	Returns set-level findings without duplicating item-level consequences.
CJMM Remediation Orchestrator	Routes atomic repairs to the exact phase, exhibit, item, response, key, feedback, or package owner.	Returns actual revised evidence to blind re-solving, affected-phase replay, set regression, and package validation.
CJMM Package and Release Validation	Reconciles the exact case, evidence, scoring, findings, remediation, accessibility, package structure, and version record.	Returns a human-review candidate only when every required case-study gate passes.

The eight Whole-Module Nursing QA lanes

Independent lane	What it asks	Where a failure returns
1 · Asset learning	Does every text and video asset perform its assigned instructional job and support the objective at the required level?	Owning text or video skill, then local checks, Nursing Module Audit when affected, and Whole-Module Nursing QA regression.
2 · H5P/NGN validity	Is every embedded interaction individually valid, necessary, evidence-grounded, and aligned?	QA Batch for Assessments and the level-specific H5P/NGN designer.
3 · L3–L5 option validity	Do higher-level response options support one defensible answer without ambiguity, cueing, or lower-level shortcuts?	Owning interaction or assessment writer and every affected answer/feedback dependency.
4 · Assessment-set distribution	Does the full assessment set cover objectives and cognitive jobs without overuse, gaps, repeated scenarios, or repeated answer patterns?	Assessment batch and the exact items needed to repair the set.
5 · Visual production	Are required visuals present, readable, accessible, accurate, instructionally purposeful, synchronized, and production-ready?	Description, image, segmentation, scene, graphics, text, or video owner depending on root cause.
6 · Clinical language	Is the module accurate, safe, within scope, evidence-supported, terminology-consistent, and clear about timing, escalation, and reassessment?	Whole-Module Clinical Language findings route to every affected asset owner and shared standard.
7 · Progression	Do L2 foundations advance through L3 application, L4 analysis, and L5 evaluation without gaps or higher levels reteaching from scratch?	The earliest upstream asset causing the gap, plus all dependent pages and assessments.
8 · Cross-asset redundancy	Does each text, video, example, H5P, and NGN add a distinct learning job rather than repeating content without purpose?	Owning asset skills; preserve essential reinforcement while removing or reframing unnecessary duplication.

Not everything in the ecosystem is a skill

Human authority	Defines intent, resolves conflicts, approves entry to later layers, and accepts the final exact version.
Source evidence	Constrains what the system may teach, key, claim, or infer.
Artifact	A page, script, interaction, bank, image, worksheet, workbook, manifest, or package that carries state between workflows.
Gate	A pass, fail, or blocked decision that controls whether the exact version advances.
Version record	Binds evidence, scores, findings, repairs, and approvals to the files that were actually checked.

The status words are not interchangeable

Draft	Still being created or revised.
Authoring-ready	Passed local skill gates; eligible for first human review.
Nursing Module Audit-verified	Exact module reached Nursing Module Audit 100; eligible for Lead review.
Lead-approved	Eligible for independent Whole-Module Nursing QA.
Whole-Module Nursing QA-verified	All applicable domains reached 100 and release validation passed; eligible for CEO review.
CEO-accepted	Final for the intended use until any later change creates a new governed version.

Complete-system rule: A skill can create, compare, audit, route, repair, validate, or package—but no single skill owns the entire truth. Release requires agreement among the exact source, the asset-specific skills, shared standards, evidence records, module-level audits, regression checks, and the authorized humans.