Statement index

Every competence statement in the framework, addressable and citable. A statement identifier is a permanent address: it is never reused, never renumbered, and resolves for as long as the framework exists.

255 statements entered · eight domains, three overlays, five extensions · version 1.29 · identifiers permanent

255 of 255 statements shown

/framework/statements

ID	Statement	Domain	Level	Type	Status
D1.L1.01	Describe in plain terms what a generative AI system does when it produces output.	D1	L1	Knowledge	ACTIVE
D1.L1.02	State that an AI system may produce different output for the same input, and why that matters.	D1	L1	Knowledge	ACTIVE
D1.L1.03	Identify tasks an AI system is unsuitable for.	D1	L1	Judgement	ACTIVE
D1.L1.04	Describe the difference between a model, an application built on it, and the service that hosts it.	D1	L1	Knowledge	ACTIVE
D1.L2.01	Select an appropriate system for a defined task from those sanctioned for use.	D1	L2	Skill	ACTIVE
D1.L2.02	State the limits of a system’s knowledge, including what it cannot access.	D1	L2	Skill	ACTIVE
D1.L2.03	Explain accurately to a colleague what a system did and did not do.	D1	L2	Skill	ACTIVE
D1.L2.04	Identify where the confidence of an output is unrelated to its correctness.	D1	L2	Skill	ACTIVE
D1.L2.05	Record which system, version and configuration produced a given output.	D1	L2	Skill	ACTIVE
D1.L2.06	Recognise a change in capability or behaviour following a model or version update.	D1	L2	Skill	ACTIVE
D1.L3.01	Assess whether a class of task lies within the capability of current systems.	D1	L3	Judgement	ACTIVE
D1.L3.02	Compare candidate models against the stated requirements of a defined task.	D1	L3	Skill	ACTIVE
D1.L3.03	Explain the mechanism behind a failure mode to a non-specialist audience.	D1	L3	Skill	ACTIVE
D1.L3.04	Assess how context length, retrieval and tool access change what a system can do.	D1	L3	Judgement	ACTIVE
D1.L3.05	Distinguish limits inherent to the approach from limits of a particular implementation.	D1	L3	Judgement	ACTIVE
D1.L3.06	Advise on the capability, cost and latency trade-off in system selection.	D1	L3	Judgement	ACTIVE
D1.L3.07	Evaluate a vendor’s description of a system against its observable behaviour.	D1	L3	Judgement	ACTIVE
D1.L3.08	Maintain currency as capability changes, and revise advice already given.	D1	L3	Practice	ACTIVE
D1.L4.01	Establish which classes of work the organisation will and will not apply AI to.	D1	L4	Judgement	ACTIVE
D1.L4.02	Set the organisation’s position on model selection, hosting and dependence.	D1	L4	Practice	ACTIVE
D1.L4.03	Establish how capability change is tracked and how existing commitments are revised.	D1	L4	Practice	ACTIVE
D1.L4.04	Advise a governing body on capability claims, including where a claim cannot be substantiated.	D1	L4	Judgement	ACTIVE
D2.L1.01	State where the information an AI system uses comes from.	D2	L1	Knowledge	ACTIVE
D2.L1.02	Recognise that information entered into a system may travel beyond the immediate task.	D2	L1	Judgement	ACTIVE
D2.L1.03	Identify information that requires a source before it is used.	D2	L1	Judgement	ACTIVE
D2.L1.04	Describe why an AI system may present unsourced content as though it were sourced.	D2	L1	Knowledge	ACTIVE
D2.L2.01	Check a factual claim in AI output against an authoritative source.	D2	L2	Skill	ACTIVE
D2.L2.02	Cite the sources used in and produced by AI-assisted work.	D2	L2	Skill	ACTIVE
D2.L2.03	Apply classification rules to material before using it with an AI system.	D2	L2	Practice	ACTIVE
D2.L2.04	Establish the licence and permitted use of material supplied to or produced by a system.	D2	L2	Skill	ACTIVE
D2.L2.05	Prepare source material so that a system can use it reliably.	D2	L2	Skill	ACTIVE
D2.L2.06	Record the provenance of AI-assisted output, including the material supplied.	D2	L2	Skill	ACTIVE
D2.L2.07	Recognise output shaped by an unrepresentative or incomplete source set.	D2	L2	Skill	ACTIVE
D2.L3.01	Design the information supply for an AI system, including sources, refresh and exclusions.	D2	L3	Skill	ACTIVE
D2.L3.02	Assess whether a source set is fit for the questions it will be asked.	D2	L3	Judgement	ACTIVE
D2.L3.03	Diagnose failures caused by source quality rather than by the model.	D2	L3	Skill	ACTIVE
D2.L3.04	Design retention, minimisation and deletion for AI workflows.	D2	L3	Skill	ACTIVE
D2.L3.05	Assess bias arising from source composition and state its consequence.	D2	L3	Judgement	ACTIVE
D2.L3.06	Establish provenance requirements for output that will be published or relied upon.	D2	L3	Skill	ACTIVE
D2.L3.07	Advise on the use of personal, confidential or licensed material with AI systems.	D2	L3	Judgement	ACTIVE
D2.L3.08	Design how conflicting sources are resolved and how the resolution is recorded.	D2	L3	Skill	ACTIVE
D2.L4.01	Establish the organisation’s standard for information used by AI systems.	D2	L4	Practice	ACTIVE
D2.L4.02	Define ownership and accountability for source sets and their maintenance.	D2	L4	Practice	ACTIVE
D2.L4.03	Set the organisation’s position on training, retention and reuse by third-party providers.	D2	L4	Judgement	ACTIVE
D2.L4.04	Hold accountable those responsible for source quality where output has caused harm.	D2	L4	Practice	ACTIVE
D3.L1.01	Describe what an instruction to an AI system needs to contain to be usable.	D3	L1	Knowledge	ACTIVE
D3.L1.02	State that the wording of an instruction changes the output.	D3	L1	Knowledge	ACTIVE
D3.L1.03	Recognise when an output has answered a different question from the one asked.	D3	L1	Judgement	ACTIVE
D3.L1.04	State that a system will answer rather than decline when a request is unclear.	D3	L1	Knowledge	ACTIVE
D3.L2.01	Write an instruction stating task, context, format and constraints.	D3	L2	Skill	ACTIVE
D3.L2.02	Supply the material a system needs rather than relying on its recall.	D3	L2	Skill	ACTIVE
D3.L2.03	Revise an instruction in response to a specific defect in the output.	D3	L2	Skill	ACTIVE
D3.L2.04	Use examples to establish the required form of an output.	D3	L2	Skill	ACTIVE
D3.L2.05	Recognise and interrupt an unproductive exchange rather than continuing it.	D3	L2	Judgement	ACTIVE
D3.L2.06	Record an instruction that worked so that it can be reused and reviewed.	D3	L2	Skill	ACTIVE
D3.L3.01	Design reusable instruction patterns for a recurring task.	D3	L3	Skill	ACTIVE
D3.L3.02	Decompose a task into steps a system can perform reliably.	D3	L3	Skill	ACTIVE
D3.L3.03	Diagnose whether a failure lies in the instruction, the material or the system.	D3	L3	Judgement	ACTIVE
D3.L3.04	Design instructions that fail safely when input falls outside their scope.	D3	L3	Skill	ACTIVE
D3.L3.05	Evaluate a change to an instruction against a case set rather than by impression.	D3	L3	Skill	ACTIVE
D3.L3.06	Advise others on interaction practice, giving reasons rather than recipes.	D3	L3	Judgement	ACTIVE
D3.L4.01	Establish organisational practice for instruction design, reuse and review.	D3	L4	Practice	ACTIVE
D3.L4.02	Define where interaction is standardised and where practitioner judgement governs.	D3	L4	Judgement	ACTIVE
D3.L4.03	Establish how instruction assets are owned, versioned and retired.	D3	L4	Practice	ACTIVE
D4.L1.01	Describe what changes when an AI step runs without a person present.	D4	L1	Knowledge	ACTIVE
D4.L1.02	Identify the steps in a workflow they perform that could run unattended.	D4	L1	Judgement	ACTIVE
D4.L1.03	State why an automated step needs a record of what it did.	D4	L1	Knowledge	ACTIVE
D4.L1.04	Recognise that a system connected to other systems can act, not only answer.	D4	L1	Knowledge	ACTIVE
D4.L2.01	Operate an automated AI workflow within its stated limits.	D4	L2	Practice	ACTIVE
D4.L2.02	Configure the handover between an AI step and the step that follows it.	D4	L2	Skill	ACTIVE
D4.L2.03	Record what an automated run did, on what input, with what result.	D4	L2	Skill	ACTIVE
D4.L2.04	Identify and report a failure in an unattended step.	D4	L2	Skill	ACTIVE
D4.L2.05	Apply the organisation’s rules on which actions an automated step may take.	D4	L2	Practice	ACTIVE
D4.L2.06	Test an automated workflow before it runs on real work.	D4	L2	Skill	ACTIVE
D4.L3.01	Design an automated AI workflow, stating where human judgement is required.	D4	L3	Skill	ACTIVE
D4.L3.02	Design error handling for an unattended AI step.	D4	L3	Skill	ACTIVE
D4.L3.03	Design the observability of an automated workflow.	D4	L3	Skill	ACTIVE
D4.L3.04	Assess the blast radius of an automated step and constrain it.	D4	L3	Judgement	ACTIVE
D4.L3.05	Design an AI system that acts across multiple steps, with authorisation boundaries.	D4	L3	Skill	ACTIVE
D4.L3.06	Diagnose a fault in a multi-step automated workflow.	D4	L3	Skill	ACTIVE
D4.L3.07	Evaluate whether automation is warranted for a given process.	D4	L3	Judgement	ACTIVE
D4.L3.08	Design the rollback of an automated workflow and its effects.	D4	L3	Skill	ACTIVE
D4.L4.01	Establish the conditions under which AI systems may act unattended in the organisation.	D4	L4	Practice	ACTIVE
D4.L4.02	Define accountability for the output of automated AI workflows.	D4	L4	Practice	ACTIVE
D4.L4.03	Establish operational standards for monitoring, alerting and incident response in AI operations.	D4	L4	Practice	ACTIVE
D4.L4.04	Govern the accumulation of automation, including workflows nobody owns.	D4	L4	Judgement	ACTIVE
D5.L1.01	State that not every problem is improved by adding an AI system.	D5	L1	Judgement	ACTIVE
D5.L1.02	Describe what a requirement is, and how it differs from a preference.	D5	L1	Knowledge	ACTIVE
D5.L1.03	Identify who is affected by a system before it is built.	D5	L1	Judgement	ACTIVE
D5.L1.04	State why the cost of a system continues after it is built.	D5	L1	Knowledge	ACTIVE
D5.L2.01	Write requirements for an AI-assisted task that can be checked.	D5	L2	Skill	ACTIVE
D5.L2.02	Identify the constraints a solution must work within.	D5	L2	Skill	ACTIVE
D5.L2.03	Describe an existing process accurately before proposing to change it.	D5	L2	Skill	ACTIVE
D5.L2.04	Compare a proposed AI solution against a non-AI alternative.	D5	L2	Judgement	ACTIVE
D5.L2.05	Identify the evidence a decision to build would require.	D5	L2	Skill	ACTIVE
D5.L2.06	Document a design decision and the reason for it.	D5	L2	Skill	ACTIVE
D5.L3.01	Decide whether to build, buy, adapt or decline.	D5	L3	Judgement	ACTIVE
D5.L3.02	Design the architecture of an AI-enabled solution, stating its boundaries.	D5	L3	Skill	ACTIVE
D5.L3.03	Specify the human role in a system before specifying its automation.	D5	L3	Skill	ACTIVE
D5.L3.04	Assess total cost of ownership, including evaluation, monitoring and change.	D5	L3	Judgement	ACTIVE
D5.L3.05	Design for failure, degradation and unavailability.	D5	L3	Skill	ACTIVE
D5.L3.06	Establish the acceptance criteria a solution must meet before release.	D5	L3	Skill	ACTIVE
D5.L3.07	Assess the effect of a solution on the people who do the work.	D5	L3	Judgement	ACTIVE
D5.L3.08	Advise on a proposal that should not proceed, and state why.	D5	L3	Judgement	ACTIVE
D5.L4.01	Set the organisation’s approach to designing and approving AI-enabled work.	D5	L4	Practice	ACTIVE
D5.L4.02	Establish which architectural dependencies the organisation will accept.	D5	L4	Judgement	ACTIVE
D5.L4.03	Decide between competing AI investments on stated criteria.	D5	L4	Judgement	ACTIVE
D5.L4.04	Establish when a system is retired, and act on it.	D5	L4	Practice	ACTIVE
D6.L1.01	Describe why AI output requires evaluation before it is relied upon.	D6	L1	Knowledge	ACTIVE
D6.L1.02	Distinguish a system that has been demonstrated from one that has been evaluated.	D6	L1	Knowledge	ACTIVE
D6.L1.03	State what a test case is and why evaluation requires more than a small number of examples.	D6	L1	Knowledge	ACTIVE
D6.L1.04	Recognise that a published benchmark result does not establish fitness for a specific task.	D6	L1	Judgement	ACTIVE
D6.L2.01	Apply defined acceptance criteria to judge whether an AI output meets a stated requirement.	D6	L2	Skill	ACTIVE
D6.L2.02	Produce a set of test cases covering expected, edge and failure conditions for a defined task.	D6	L2	Skill	ACTIVE
D6.L2.03	Document evaluation results so that another person can reproduce the judgement.	D6	L2	Skill	ACTIVE
D6.L2.04	Check whether a change to a prompt, model or configuration has altered previously acceptable output.	D6	L2	Skill	ACTIVE
D6.L2.05	Record an observed failure with sufficient context for it to be reproduced and investigated.	D6	L2	Skill	ACTIVE
D6.L2.06	Report evaluation findings, distinguishing measured results from impressions.	D6	L2	Skill	ACTIVE
D6.L3.01	Design fitness criteria for an AI system before development, derived from its intended use.	D6	L3	Skill	ACTIVE
D6.L3.02	Design an evaluation set representative of production conditions and resistant to overfitting.	D6	L3	Skill	ACTIVE
D6.L3.03	Evaluate agreement between human raters and diagnose the causes of disagreement.	D6	L3	Skill	ACTIVE
D6.L3.04	Justify the choice of automated, model-graded or human evaluation for a given task.	D6	L3	Judgement	ACTIVE
D6.L3.05	Design regression testing that detects degradation across model and configuration changes.	D6	L3	Skill	ACTIVE
D6.L3.06	Critique a published benchmark or vendor claim, identifying what it does and does not establish.	D6	L3	Judgement	ACTIVE
D6.L3.07	Design production monitoring that detects drift, degradation and emerging failure modes.	D6	L3	Skill	ACTIVE
D6.L3.08	Review an incident involving AI output and identify the assurance gap that allowed it.	D6	L3	Judgement	ACTIVE
D6.L4.01	Establish evaluation standards and evidence requirements applying across an organisation’s AI systems.	D6	L4	Practice	ACTIVE
D6.L4.02	Define the thresholds at which a system may enter, remain in, or must be withdrawn from use.	D6	L4	Judgement	ACTIVE
D6.L4.03	Commission independent evaluation where internal assurance is insufficient or conflicted.	D6	L4	Judgement	ACTIVE
D6.L4.04	Hold accountable those responsible for assurance evidence, including where evidence is absent.	D6	L4	Practice	ACTIVE
D7.L1.01	State that using an AI system does not transfer responsibility for the result.	D7	L1	Knowledge	ACTIVE
D7.L1.02	Identify where AI use must be disclosed.	D7	L1	Judgement	ACTIVE
D7.L1.03	Recognise that AI output can affect people unequally.	D7	L1	Knowledge	ACTIVE
D7.L1.04	State the obligations that apply to their own AI use.	D7	L1	Knowledge	ACTIVE
D7.L2.01	Disclose AI involvement in work where disclosure is required.	D7	L2	Practice	ACTIVE
D7.L2.02	Apply organisational and legal obligations to a specific AI task.	D7	L2	Practice	ACTIVE
D7.L2.03	Check AI-assisted work for effects that fall unevenly on a group of people.	D7	L2	Skill	ACTIVE
D7.L2.04	Attribute and respect the rights in material used and produced.	D7	L2	Practice	ACTIVE
D7.L2.05	Raise a concern about an AI use they believe to be wrong.	D7	L2	Practice	ACTIVE
D7.L2.06	Explain to an affected person, accurately, how AI was involved in a decision.	D7	L2	Skill	ACTIVE
D7.L3.01	Assess the impact of a proposed AI use on the people it affects.	D7	L3	Judgement	ACTIVE
D7.L3.02	Design fairness requirements for a system and the evidence they require.	D7	L3	Skill	ACTIVE
D7.L3.03	Establish transparency and explanation requirements for a system.	D7	L3	Skill	ACTIVE
D7.L3.04	Advise on a use that is permitted but should not proceed.	D7	L3	Judgement	ACTIVE
D7.L3.05	Design the route by which an affected person can challenge an AI-influenced decision.	D7	L3	Skill	ACTIVE
D7.L3.06	Assess the labour and skill effects of an AI deployment.	D7	L3	Judgement	ACTIVE
D7.L3.07	Apply the obligations of a regulated or professional context to AI use.	D7	L3	Judgement	ACTIVE
D7.L3.08	Review an AI-related incident for the ethical failure, not only the technical one.	D7	L3	Judgement	ACTIVE
D7.L4.01	Establish the organisation’s principles for AI use, in terms that decide cases.	D7	L4	Practice	ACTIVE
D7.L4.02	Define accountability for AI-influenced decisions across the organisation.	D7	L4	Practice	ACTIVE
D7.L4.03	Establish how concerns and dissent about AI use are heard and acted on.	D7	L4	Practice	ACTIVE
D7.L4.04	Account publicly for the organisation’s AI use, including where it went wrong.	D7	L4	Judgement	ACTIVE
D8.L1.01	Describe how content an AI system reads can alter its behaviour.	D8	L1	Knowledge	ACTIVE
D8.L1.02	Identify information that must not be entered into an AI system.	D8	L1	Judgement	ACTIVE
D8.L1.03	State that AI output may be unsafe for a downstream system or person to act on.	D8	L1	Knowledge	ACTIVE
D8.L1.04	Recognise AI use falling outside sanctioned tools and channels.	D8	L1	Judgement	ACTIVE
D8.L1.05	Describe why an AI system’s access to other systems is a security concern.	D8	L1	Knowledge	ACTIVE
D8.L2.01	Follow organisational policy on which AI systems may be used, and for what material.	D8	L2	Practice	ACTIVE
D8.L2.02	Check AI output for content that should not be passed to a downstream system or person.	D8	L2	Skill	ACTIVE
D8.L2.03	Configure an AI system’s access so that it holds only the permissions its task requires.	D8	L2	Skill	ACTIVE
D8.L2.04	Apply organisational controls for secrets and credentials in AI workflows.	D8	L2	Practice	ACTIVE
D8.L2.05	Recognise and report suspected manipulation of an AI system.	D8	L2	Skill	ACTIVE
D8.L2.06	Document the data an AI workflow reads, retains and transmits.	D8	L2	Skill	ACTIVE
D8.L2.07	Check the provenance and requested access of a model, tool or connector before use.	D8	L2	Skill	ACTIVE
D8.L3.01	Design an AI feature so that untrusted content cannot cause privileged action.	D8	L3	Skill	ACTIVE
D8.L3.02	Threat model an AI feature, identifying assets, entry points and adversaries.	D8	L3	Skill	ACTIVE
D8.L3.03	Design controls that limit what a system can do when its instructions are compromised.	D8	L3	Skill	ACTIVE
D8.L3.04	Evaluate the effectiveness of guardrails and output controls under realistic conditions.	D8	L3	Judgement	ACTIVE
D8.L3.05	Diagnose how sensitive data reached a prompt, context window, log or output.	D8	L3	Skill	ACTIVE
D8.L3.06	Evaluate supply chain risk in models, tools, extensions and connectors.	D8	L3	Judgement	ACTIVE
D8.L3.07	Design least-privilege access for systems acting on behalf of people.	D8	L3	Skill	ACTIVE
D8.L3.08	Resolve an incident in which an AI system was the vector, and identify the control that failed.	D8	L3	Judgement	ACTIVE
D8.L3.09	Advise on unsanctioned AI use, addressing the need that produced it.	D8	L3	Judgement	ACTIVE
D8.L4.01	Establish the security standard AI systems must meet before deployment.	D8	L4	Practice	ACTIVE
D8.L4.02	Define the conditions under which a system’s access or autonomy must be reduced or withdrawn.	D8	L4	Judgement	ACTIVE
D8.L4.03	Govern the introduction of AI models, tools and connectors across an organisation.	D8	L4	Practice	ACTIVE
D8.L4.04	Lead response to an incident in which an AI system caused or enabled harm.	D8	L4	Practice	ACTIVE
O1.L2.01	Deliver instruction against named statements at a named level.	O1	L2	Practice	ACTIVE
O1.L2.02	Collect evidence of a learner’s performance against stated indicators.	O1	L2	Skill	ACTIVE
O1.L2.03	Give feedback that names the indicator not yet met and what would meet it.	O1	L2	Skill	ACTIVE
O1.L3.01	Design instruction from a set of statements and a target level.	O1	L3	Skill	ACTIVE
O1.L3.02	Construct assessment tasks that produce evidence an assessor can judge.	O1	L3	Skill	ACTIVE
O1.L3.03	Judge borderline and insufficient evidence consistently.	O1	L3	Judgement	ACTIVE
O1.L3.04	Establish and monitor agreement between assessors.	O1	L3	Skill	ACTIVE
O1.L3.05	Declare and manage the limits of one’s own competence in the domains taught.	O1	L3	Practice	ACTIVE
O1.L4.01	Establish the standard for instruction and assessment across a provider.	O1	L4	Practice	ACTIVE
O1.L4.02	Govern conflict of interest where instruction and assessment are held together.	O1	L4	Judgement	ACTIVE
O1.L4.03	Account for the validity of claims a provider makes about coverage and level.	O1	L4	Practice	ACTIVE
O2.L2.01	Apply the organisation’s AI policy in day-to-day enablement work.	O2	L2	Practice	ACTIVE
O2.L2.02	Record and route requests for AI tools, access and support.	O2	L2	Skill	ACTIVE
O2.L2.03	Support colleagues in using sanctioned AI systems for their own work.	O2	L2	Practice	ACTIVE
O2.L3.01	Assess organisational readiness for a proposed AI deployment.	O2	L3	Judgement	ACTIVE
O2.L3.02	Design a competence development plan mapped to framework statements.	O2	L3	Skill	ACTIVE
O2.L3.03	Design the support model for AI use, including escalation.	O2	L3	Skill	ACTIVE
O2.L3.04	Diagnose why a sanctioned tool is not being adopted.	O2	L3	Judgement	ACTIVE
O2.L3.05	Measure the effect of AI adoption on the work, not on tool usage.	O2	L3	Judgement	ACTIVE
O2.L4.01	Establish the organisation’s AI capability strategy against defined competence.	O2	L4	Practice	ACTIVE
O2.L4.02	Define the provision that makes policy followable.	O2	L4	Judgement	ACTIVE
O2.L4.03	Account to the governing body for AI capability and its gaps.	O2	L4	Practice	ACTIVE
O3.L2.01	Collect evidence for an AI assurance review against a defined scope.	O3	L2	Skill	ACTIVE
O3.L2.02	Test a stated control and record whether it operates as described.	O3	L2	Skill	ACTIVE
O3.L2.03	Record a finding so that it can be acted on and re-tested.	O3	L2	Skill	ACTIVE
O3.L3.01	Design the scope of an AI assurance review.	O3	L3	Skill	ACTIVE
O3.L3.02	Evaluate whether evidence supports the claim being made.	O3	L3	Judgement	ACTIVE
O3.L3.03	Assess a coverage map or conformance claim critically.	O3	L3	Judgement	ACTIVE
O3.L3.04	Maintain independence and manage conflict in an assurance engagement.	O3	L3	Practice	ACTIVE
O3.L3.05	Report assurance conclusions, including unwelcome ones, with their basis.	O3	L3	Skill	ACTIVE
O3.L4.01	Establish an assurance programme for AI systems across an organisation.	O3	L4	Practice	ACTIVE
O3.L4.02	Commission and evaluate independent assurance.	O3	L4	Judgement	ACTIVE
O3.L4.03	Hold the organisation to the standard its assurance claims imply.	O3	L4	Practice	ACTIVE
X-AGT.L2.01	Operate an agentic system within a defined task boundary.	X-AGT	L2	Practice	ACTIVE
X-AGT.L2.02	Review an agent’s action log to establish what it actually did.	X-AGT	L2	Skill	ACTIVE
X-AGT.L2.03	Apply authorisation rules to actions an agent may take unattended.	X-AGT	L2	Practice	ACTIVE
X-AGT.L3.01	Design an agent’s tool access and authorisation boundaries.	X-AGT	L3	Skill	ACTIVE
X-AGT.L3.02	Design termination, budget and loop conditions for an agentic task.	X-AGT	L3	Skill	ACTIVE
X-AGT.L3.03	Evaluate an agentic system across multi-step trajectories, not single outputs.	X-AGT	L3	Skill	ACTIVE
X-AGT.L3.04	Design containment for an agent whose instructions have been compromised.	X-AGT	L3	Skill	ACTIVE
X-AGT.L4.01	Establish the conditions under which agentic systems may operate in the organisation.	X-AGT	L4	Practice	ACTIVE
X-AGT.L4.02	Define accountability for actions taken by an agent.	X-AGT	L4	Practice	ACTIVE
X-RAG.L2.01	Establish which sources a grounded system can and cannot reach.	X-RAG	L2	Skill	ACTIVE
X-RAG.L2.02	Verify that a grounded answer is supported by the passage it cites.	X-RAG	L2	Skill	ACTIVE
X-RAG.L2.03	Report retrieval failures distinctly from generation failures.	X-RAG	L2	Skill	ACTIVE
X-RAG.L3.01	Design the index, chunking and refresh strategy for a grounded system.	X-RAG	L3	Skill	ACTIVE
X-RAG.L3.02	Evaluate retrieval quality separately from answer quality.	X-RAG	L3	Skill	ACTIVE
X-RAG.L3.03	Design grounding requirements and behaviour when grounding is absent.	X-RAG	L3	Skill	ACTIVE
X-RAG.L3.04	Design access control so retrieval respects the requester’s entitlements.	X-RAG	L3	Skill	ACTIVE
X-RAG.L4.01	Establish standards for grounded systems, including provenance and currency.	X-RAG	L4	Practice	ACTIVE
X-RAG.L4.02	Govern the corpora on which grounded systems depend.	X-RAG	L4	Judgement	ACTIVE
X-MOD.L2.01	Document the data, configuration and lineage of a training or tuning run.	X-MOD	L2	Skill	ACTIVE
X-MOD.L2.02	Apply data handling and licensing rules to training material.	X-MOD	L2	Practice	ACTIVE
X-MOD.L2.03	Evaluate a tuned model against a held-out set before release.	X-MOD	L2	Skill	ACTIVE
X-MOD.L3.01	Design a dataset for a stated objective, including its exclusions.	X-MOD	L3	Skill	ACTIVE
X-MOD.L3.02	Justify the choice between prompting, retrieval, tuning and training.	X-MOD	L3	Judgement	ACTIVE
X-MOD.L3.03	Evaluate a model for regression, bias and safety behaviour before release.	X-MOD	L3	Skill	ACTIVE
X-MOD.L3.04	Document a model so that others can judge its fitness.	X-MOD	L3	Skill	ACTIVE
X-MOD.L4.01	Establish standards for model development, evaluation and release.	X-MOD	L4	Practice	ACTIVE
X-MOD.L4.02	Govern the data on which models are trained, and account for it.	X-MOD	L4	Judgement	ACTIVE
X-GEN.L2.01	Disclose and label synthetic media as the context requires.	X-GEN	L2	Practice	ACTIVE
X-GEN.L2.02	Establish rights in generated media and the material it derives from.	X-GEN	L2	Skill	ACTIVE
X-GEN.L2.03	Review generated media for harmful, misleading or unintended content.	X-GEN	L2	Skill	ACTIVE
X-GEN.L3.01	Design review and approval for generated media before publication.	X-GEN	L3	Skill	ACTIVE
X-GEN.L3.02	Assess representation and stereotype effects in generated output at scale.	X-GEN	L3	Judgement	ACTIVE
X-GEN.L3.03	Design provenance and disclosure for generated assets across their lifecycle.	X-GEN	L3	Skill	ACTIVE
X-GEN.L3.04	Advise on a generated asset that should not be published.	X-GEN	L3	Judgement	ACTIVE
X-GEN.L4.01	Establish the organisation’s position on synthetic media and its disclosure.	X-GEN	L4	Practice	ACTIVE
X-GEN.L4.02	Account for the organisation’s use of synthetic media, including failures.	X-GEN	L4	Judgement	ACTIVE
X-SWE.L2.01	Review AI-generated code before it enters a codebase.	X-SWE	L2	Skill	ACTIVE
X-SWE.L2.02	Test AI-generated code, including cases it was not written for.	X-SWE	L2	Skill	ACTIVE
X-SWE.L2.03	Establish the provenance and licence of generated or suggested code.	X-SWE	L2	Practice	ACTIVE
X-SWE.L3.01	Design review practice for AI-assisted development.	X-SWE	L3	Skill	ACTIVE
X-SWE.L3.02	Assess the security implications of AI-generated code and dependencies.	X-SWE	L3	Judgement	ACTIVE
X-SWE.L3.03	Design the boundary between AI-assisted and human-authored responsibility.	X-SWE	L3	Skill	ACTIVE
X-SWE.L3.04	Assess the effect of AI assistance on maintainability and team competence.	X-SWE	L3	Judgement	ACTIVE
X-SWE.L4.01	Establish engineering standards for AI-assisted development.	X-SWE	L4	Practice	ACTIVE
X-SWE.L4.02	Account for the quality and security of AI-assisted output in production.	X-SWE	L4	Practice	ACTIVE

Coverage today

`D1`	AI foundations	22
`D2`	Data and information	23
`D3`	Interaction and elicitation	19
`D4`	Automation and integration	22
`D5`	Solution design	22
`D6`	Evaluation and assurance	22
`D7`	Ethics, governance and professional practice	22
`D8`	Security	25
`O1`		11
`O2`		11
`O3`		11
`X-AGT`		9
`X-RAG`		9
`X-MOD`		9
`X-GEN`		9
`X-SWE`		9

All eight domains, three overlays and five extensions are entered at version 1.29. Identifiers are permanent and are never reused, so a citation made today resolves to the same statement in every later release.

How to cite a filtered view

Filters are carried in the address, so any view of the index can be cited as it was seen.

/framework/statements?domain=D6&level=L3

To cite a fixed release rather than the current view, cite the release record: /framework/v1.29.

This page displays version 1.29 · last reviewed 30.08.2026