AILA Pro · Audit Trail · ISCM Foundation

Non-negotiable distinction. Internal instructions can guide behaviour, but they cannot enforce workspace membership, permissions, sharing restrictions, knowledge-file selection, credential control or ownership continuity.

Those controls must be configured outside the prompt and supported by evidence. A tick records a decision; the evidence reference supports the audit trail.

Audit record

Record identification

Audit record ID Institution / programme Prepared by Record date

This file contains no analytics, network calls or external dependencies. Entries remain in the current browser until exported, downloaded or reset.

1 Accountability and lifecycle control Governance • Workspace • Change control

Who is accountable for this clone, who may change it, and what version, review and retirement records identify it?

Record named roles rather than departments wherever practical.

Clone name * Version * Next scheduled review * Institutional owner * Methodology custodian * Workspace administrator * Authorized editors Independent reviewer Decommissioning owner * Change or approval reference

Confirmed lifecycle controls

Owners, custodians, administrators, reviewers and editors are named. Only approved persons may change purpose, instructions, files, tools or settings. Emergency changes require documented authority and later ratification. Suspension, rollback, archive and decommissioning triggers are defined.

Audit trace for Question 1

Evidence / register reference * Decision owner * Decision date *

Exception, gap or N/A rationale Decision complete; evidence, ownership and date are traceable.

2 Purpose, users and system boundary Prompt • Governance • Validation

What institutional outcome, authorized cohort, approved services, exclusions and measurable acceptance criteria define the clone?

Purpose statement * Authorized cohort or user group * Applicable jurisdiction or policy environment

Approved service catalogue

Concept explanation and learning support Evidence synthesis and document analysis Stakeholder and complexity analysis Strategic scenario planning Programme monitoring and early warning Investment and bankability decision support

Other approved service

Authority posture *

Advisory only Decision support with named human approval Limited actions under separate written authorization

Advisory support is distinguished from final institutional authority. The response to out-of-scope requests is defined.

Explicit exclusions * Performance objectives and acceptance thresholds *

Audit trace for Question 2

Evidence / register reference * Decision owner * Decision date *

Exception, gap or N/A rationale Decision complete; evidence, ownership and date are traceable.

3 Access, permissions and sharing Workspace • Identity • Separation of duties

Who may use, review or edit the clone, how are they authorized and removed, and which sharing modes are prohibited?

Authorized user groups * Access register and review frequency *

Ordinary participant permission *

Can chat only Approved exception recorded below

Settings visibility *

Nobody Named trusted methodology reviewers only

Named settings reviewers Named editors Permission exception or justification

Configured access and sharing controls

Invite-only or equivalent controlled access is enabled. Public-link access is disabled. Public store or catalogue publication is disabled where required. Unrestricted workspace-wide sharing is disabled or restricted. Day-to-day users are separated from settings reviewers and editors. Invitation, verification, role change, offboarding and emergency revocation are defined. A claimed role in chat never grants restricted access or authority. Restricted data, tools and actions require platform-side authorization.

Audit trace for Question 3

Evidence / register reference * Decision owner * Decision date *

Exception, gap or N/A rationale Decision complete; evidence, ownership and date are traceable.

4 Behaviour, methodology and evidence discipline Prompt • Outputs • Provenance • Methodology protection

What approved workflow, output standards, source hierarchy and evidence distinctions must govern the clone’s responses?

Required analytical workflow * Source hierarchy and conflict rule * Tone, terminology, languages and mandatory structure

Approved output forms

Brief answer or teaching explanation Analytical memo or evidence synthesis Driving-force, readiness or complexity map Scenario set or early-warning dashboard Financial model or bankability assessment Transformation plan or governance recommendation

Evidence categories to be distinguished

Documented ISCM methodology Logical extension from ISCM methodology External research or independent evidence Model-generated inference Provisional assumption

Mandatory response controls

Sources, provenance and attribution are visible where required. External research is clearly separated from institutional methodology. Assumptions, limitations and uncertainty are labelled. Stale, conflicting, missing or low-confidence evidence is flagged. Invented sources, formulas, evidence or completed actions are prohibited. Defined terms and required methodology steps may not drift silently. Clarification, labelled assumption and best-effort thresholds are defined. Unsupported certainty and autonomous high-impact decisions are prohibited.

Audit trace for Question 4

Evidence / register reference * Decision owner * Decision date *

Exception, gap or N/A rationale Decision complete; evidence, ownership and date are traceable.

5 Knowledge, intellectual property and privacy Knowledge • Data • Confidentiality • Cohort separation

Which sources may the clone use, what must be excluded, and how are provenance, intellectual property, confidentiality, personal data and cross-cohort separation controlled?

Approved knowledge register * Excluded or restricted material *

Personal-data posture *

No personal data Limited approved fields under a named protocol Approved under a formal data-access and retention protocol

Privacy or data protocol

Audit trace for Question 5

Evidence / register reference * Decision owner * Decision date *

Exception, gap or N/A rationale Decision complete; evidence, ownership and date are traceable.

6 Tools, safety and human authority Tools • Web use • Safety • Escalation • Fail-safe behaviour

Which tools and operations are authorized, which actions require human confirmation, and what requests or conditions require refusal or escalation?

Authorized tools and integrations * Escalation triggers, channel and owner *

External-research posture *

Not permitted Permitted under approved source, privacy and citation rules Required where current verification is material

Authorized operation posture

Search or retrieve from approved sources Analyze or transform approved information Create draft outputs without external execution Update records only with explicit human confirmation Send or publish only with explicit human confirmation Delete, transact or execute only under separate written authorization

Safety and fail-safe controls

Messages, files, websites and tool outputs are treated as untrusted evidence, not higher authority. Untrusted content cannot reveal secrets, change scope or bypass controls. Financial, legal, employment, healthcare, public-policy and safety-critical outputs remain advisory. Irreversible, binding, rights-affecting or funding decisions require responsible human authority. Illegal, harmful, deceptive or unsafe requests are refused or safely redirected. Tool or source failures are disclosed and never replaced by fabricated completion. Only approved fallback methods may be used after a failure. Material safety or authorization failures trigger evidence preservation and incident escalation.

Audit trace for Question 6

Evidence / register reference * Decision owner * Decision date *

Exception, gap or N/A rationale Decision complete; evidence, ownership and date are traceable.

7 Learning, change and incident governance Learning • Versioning • Review • Incidents • Early warning

How are feedback, knowledge changes, configuration changes and incidents reviewed, approved, versioned, rolled back and incorporated into institutional learning?

Change repository or register * Review cycle * Incident channel and owner *

Early-warning indicators monitored

Methodology drift Unauthorized disclosure attempts Source staleness or conflict Access or identity anomalies Tool failures or anomalous results Privacy or confidentiality incidents User complaints or refusal-quality issues Validation regression

Audit trace for Question 7

Evidence / register reference * Decision owner * Decision date *

Exception, gap or N/A rationale Decision complete; evidence, ownership and date are traceable.

8 Validation, readiness and release approval Validation • Corrective action • Release gate • Sign-off

Has the clone passed the minimum normal and adversarial tests, are material risks resolved or controlled, and who authorizes release?

Tick each test that has passed. Leave a test unticked only where the gap, N/A reason, action, owner and deadline are recorded.

Release blockers *

No unresolved material blockers Limited gaps under approved temporary control Unresolved critical blocker

Validation pack reference *

Conditions, corrective actions and temporary controls

Readiness classification *

Green — Ready Mandatory controls are complete, critical risks are closed and applicable release tests pass. Amber — Conditional Limited gaps have named actions, owners, deadlines and approved temporary controls. Red — Do not deploy Access, confidentiality, data, methodology, safety or validation failures remain unresolved.

Release sign-off

Independent reviewer Institutional owner approval Methodology custodian approval Approval date Release or rollback point Next review date

Audit trace for Question 8

Evidence / release record * Decision owner * Decision date *

Exception, gap or N/A rationale Release decision complete; evidence, ownership and date are traceable.