#U.ActionInvitationPrecisionRestoration — Affordance / Action-Invitation Precision Restoration (ACT-INV)

Pattern A.6.A · Stable · Architectural (A) · Normative (Core / Draft) Part A - Kernel Architecture Cluster

Type: Architectural (A) Status: Draft Normativity: Normative (Core / Draft)

Plain-name. Affordance / action-invitation precision restoration.

Intent. Provide a reusable discipline for repairing overloaded affordance / action-first language in FPF texts.

This pattern is an A.6.P RPR specialisation for post-threshold material: it turns bare action-oriented prose into one explicit, slot-explicit action invitation relation family with a declared sense family, lawful normal forms (CuePack | ActionOption | OptionSet | PolicyHook), explicit change semantics, and lexical guardrails. Pre-threshold action-guiding material remains with A.16.1 / B.4.1 until the cue is articulated enough for actionInvitation(...) publication. It does not mint a parallel execution ontology: whenever an invitation is articulated far enough to reference executable artifacts, publication SHALL dock to A.15 surfaces (U.Method, U.MethodDescription, U.WorkPlan, and later U.Work) rather than inventing new action kinds by prose.

It allows ecological-psychology, phenomenological, active-inference, control-theoretic, interface, engineering-operations, and robotics uses to coexist without false identity by label.

Placement. Part A > cluster A.6 Signature Stack & Boundary Discipline > specialisation of A.6.P for under-specified affordance / action-first language.

Builds on. A.3, A.6, A.6.B, A.6.P, A.6.S, A.6.0, A.6.5, A.2.6, A.7, A.15, E.8, E.10, F.9, F.18.

Coordinates with. A.6.Q for evaluative-language repair; C.2.2a / A.16 / A.16.1 / A.16.2 / B.4.1 for language-state chart positions, articulation/closure coordination, lawful moves, early cue routing, responsibility handoff, and lawful retreat when a published invitation must be reopened; use A.16.0 only when lineage, branch, loss, or handoff history itself must be published as an explicit trajectory account; B.5.2.0 when the strongest lawful continuation is still an open probe question rather than an invitation; C.2.LS / C.2.4 / C.2.5 / C.2.6 / C.2.7 for articulation, closure, anchoring, and representation-factor facets referenced but not owned here; A.10/B.3 for evidence and assurance; B.4/B.5 for anomaly-driven cycles; E.17/E.18 for viewpoint publication; F.9.1 for bridge-stance annotations; C.3.3 for kind-bridge repair when endpoint kind mismatches appear.

Non-goal. This pattern does not assert that physical affordances, interface affordances, social affordances, epistemic probe moves, articulation-closure moves, latent policy cues, and control opportunities are one concept.

Its job is to publish a disciplined bridge reading across those traditions while preventing false identity by shared language.

It also does not assert that every trigger use of action-first language is lawfully repaired by actionInvitation(...):

• where the repaired statement is primarily evaluative, use A.6.Q;
• where it is primarily about general capability, use functional / method description;
• where it is primarily deontic, route via A.6.B;
• where it is primarily about scheduled or executed enactment, dock to A.15 (U.Method / U.MethodDescription / U.WorkPlan, and later U.Work) rather than letting actionInvitation(...) become a shadow execution model.

FPF repeatedly encounters a predictable precision failure mode around affordance / action-first language.

#Keywords

• affordance
• action invitation
• action-first language
• post-threshold routing
• A.15 docking
• language-state seam.

#Relations

#Content

#Problem frame

FPF repeatedly encounters a predictable precision failure mode around affordance / action-first language.

Authors say:

• “this handle affords pulling”
• “the interface invites confirmation”
• “the alarm calls for rollback”
• “this discrepancy suggests probing deeper”
• “the draft is ready for formalization”
• “the model wants to brake”
• “the situation is actionable”

…but the intended meaning is actually one of several different action-oriented families, for example:

1. Physical affordance — a material/environmental configuration offers a bodily action to an embodied agent.
2. Interface affordance — an interface face, operator panel, alarm, or publication face presents an operator move.
3. Social affordance — another agent or interactional setting invites a response or coordination move.
4. Epistemic probe move — a problem situation invites asking, comparing, measuring, testing, or instrumenting.
5. Closure-advance move — a situation invites naming, rescoping, proxy declaration, or formalization.
6. Latent policy cue — a learned/distributed state carries an action-oriented tendency not yet locally articulated.
7. Control opportunity — a closed-loop state invites braking, rollback, replan, isolate, escalate, or override.

The recurrent failure modes are:

• Site confusion. The locus of the invitation is unclear: object, scene, interface object, description, carrier, policy state, or problem episode.
• Enactor confusion. It is unclear which system / collective system / role-holder is invited to act: human operator, robot controller, research team, review service, or some unnamed “system”.
• Action confusion. The candidate action is hidden behind vague language like actionable, calls for, ready for, natural next step.
• Invitation vs obligation collapse. A situation that merely invites an action is rewritten as if it already created a duty.
• Invitation vs capability collapse. A local, situated action opportunity is rewritten as if it were a general capability claim.
• Invitation vs work collapse. Offered action is narrated as if it had already been executed.
• Substrate confusion. Ecological, embodied, latent-distributed, and symbolic-local action cues are silently collapsed.
• Bridge illusion. Similar language across traditions is mistaken for sameness.
• Premature closure. An early cue is published as if it were already a committed method, gate, or policy.

#Problem

How can FPF let authors use the communicative convenience of affordance / action-first language while preventing category errors when the language crosses:

• ecological / phenomenological discourse,
• interface and operator-facing discourse,
• active-inference / world-model discourse,
• control / monitoring / incident-response discourse,
• robotics / embodied-AI discourse,
• epistemic exploration and problem-framing discourse?

#Forces

• Action speed vs auditability. Action-first language is attractive because it is fast; that same speed makes it unsafe at boundaries.
• Situated coupling vs explicit publication. Affordances arise in agent–environment or policy–world coupling, but boundary use requires explicit local publication.
• Preconceptual cue vs later articulation. Some invitations are real before they are stably worded.
• Enactor specificity vs shared discourse. A cue may be visible to one detector yet relevant to another would-be enactor.
• Opportunity vs obligation. Not every invitation is a gate or commitment.
• Option plurality vs premature scalarisation. Several candidate actions may co-exist without a lawful total ordering.
• Cross-tradition dialogue vs false unification. The framework should support parallels without asserting identity.
• Progressive closure. An action cue may later become an option, then a policy hook, and only later a formal gate or work plan.

#Solution - Stable lens -> Sense Family -> Slots -> Normal Form -> Change Lexicon -> Guardrails

#Trigger rule

A use of affordance / action-first language is in scope for A.6.A when any of the following holds:

• the prose uses tokens such as affords, invites, calls for, actionable, ready for, ripe for, natural next step, the model wants, the interface tells, this problem asks for;
• a boundary, gate, incident note, design note, or review note uses such language for admission, selection, triage, or action guidance;
• different traditions are compared using the same action-first wording;
• a draft introduces model affordance, interface affordance, actionable insight, policy invitation, or ready for formalization without declared sense;
• the author intends the phrase to carry more than one of: situational action opportunity, latent cue, operator move, probe move, closure move, or control move.

#Operational repair sequence

When the trigger fires, authors SHOULD follow the A.6.P repair path:

1. Capture the trigger span. Copy the exact surface phrase.

2. Reconstruct the candidate set. Enumerate plausible candidate interpretations, including:

   • candidate relation families (actionInvitation vs evaluativeAscription vs capability claim vs commitment vs work occurrence),
   • candidate site lane maps over A.7 (Object | Description | Carrier),
   • candidate would-be enactor lanes,
   • candidate action tuples.

   If the occurrence is decision-bearing or publication-bearing, record a short Candidate-Set Note before selecting a repair.

3. Select one explicit action-invitation sense. Pick one ActionInvitationSense token and state why rivals were rejected in this local context.

4. Emit a slot-explicit rewrite. Rewrite the sentence into one explicit actionInvitation(...) record with site, would-be enactor, candidate action, coupling frame, detector/viewpoint, normal form, and qualifiers.

5. Route boundary-bearing consequences. If the repaired statement is used for admissibility, commitments, publication, automation, or evidence-bearing decisions, route the downstream hooks through A.6.B and, where enactment is implied, through A.15, instead of letting the vague action-first phrase carry that burden by itself.

#Post-threshold lens: action-invitation routing anchored by actionInvitation(...)

A.6.A stabilises the ambiguity cluster by treating in-scope post-threshold affordance/action-first statements as qualified action-oriented material that must publish an explicit action-invitation normal form and declared downstream routing, not as bare adjectives or rhetorical verbs. Early action-guiding material may remain in A.16.1 / B.4.1 as cue-pack content, a RoutedCueSet, or another typed route-bounded upstream publication before this pattern is entered. A.6.A is therefore entered only once local AE is high enough to name site / enactor / action structure explicitly and local CD is high enough that one invitation reading is worth publishing as a relation record rather than remaining mere route pressure. If the strongest lawful publication is still a cue pack, routed cue, or open abductive prompt, stay in A.16.1 / B.4.1 / B.5.2.0. If a published actionInvitation(...) later loses that minimal articulation/closure support, retreat via A.16.2 rather than leaving a stale invitation record in force.

In A.6.P terms, this pattern fixes one post-threshold relation family and one downstream routing discipline:

• actionInvitation — the explicit post-threshold relation kind for affordance, invitation, control-opportunity, probe-move, and closure-advance rewrites once the material is articulated enough to publish a relation record.

#RelationKind contract skeleton for actionInvitation

The family-specific RelationKind token is actionInvitation. Its contract publication SHALL declare, at minimum:

• (L) applicability in the local Context/plane set;
• (L) site-centred polarity: the relation is about a site/situation inviting a candidate action for an enactor; it SHALL NOT be silently rewritten as a monadic property of an object alone;
• (L) participant SlotSpecs for site, invited enactor, candidate action, sense, coupling frame, and normal-form positions;
• (A) repair paths for site-kind and enactor-kind mismatches: explicit narrowing, KindBridge, and/or retargetSite(...) / retargetInvitedEnactor(...);
• (L) qualifier expectations for scope, Γ_time, viewpoint, view, representationSubstrate, bridgeRef, and (when relevant) articulationHint;
• (D) detector/enactor separation discipline: the perceiver or detector SHALL NOT be silently collapsed into the invited enactor when they differ;
• (D) obligation barrier: invitation language SHALL NOT be silently rewritten as duty language;
• (A/E) witness discipline for decision/publication/automation lanes;
• (L/A) admissible semantic change classes and edition-fence expectations;
• (A/E) cross-context / cross-plane policy when reuse is claimed.

Each in-scope occurrence SHALL be representable as a pattern-specific QualifiedRelationRecord:

ActionInvitationRecord := ⟨ relationKind : actionInvitation, siteTuple : …, siteFacetMap? : tuple-member ↦ (Object | Description | Carrier), invitedEnactorTuple : …, candidateActionTuple : …, actionInvitationSense : ActionInvitationSense, couplingFrame : …, detector? : …, viewpoint? : U.Viewpoint, view? : U.View, normalForm : CuePack | ActionOption | OptionSet | PolicyHook, articulationHint? : open-cue | sketched | option-explicit | hook-explicit, scope? : U.Scope, Γ_time? : U.GammaTimePolicy, representationSubstrate? : ecological-world-coupled | embodied-kinesthetic | latent-distributed | symbolic-local | hybrid, bridgeRef? : BridgeId, witnesses? : EvidenceRefSet ⟩

So the sentence “X affords Y” is never accepted as a terminal form. Within the scope of A.6.A it must be rewritten into an explicit actionInvitation(...) instance with declared downstream routing; earlier pre-threshold material may instead remain as cue-pack content, a RoutedCueSet, or another typed route-bounded upstream publication before A.6.A entry.

Discipline note. ActionInvitationSense is a slot value inside the relation family; it is not a replacement for the relation family itself. The stable intermediate lens is the actionInvitation(...) relation; the sense token refines what kind of invitation is being published.

A.7 lane note. siteFacetMap uses only the A.7 lane distinction Object | Description | Carrier. If a PublicationSurface / InteropSurface participates, declare it separately under A.7 / L-SURF rather than widening the lane set with a generic Surface token.

Separation note. detector and invitedEnactor are not synonyms. When both matter, they SHALL be published separately.

Enactor note. When invitedEnactorTuple is published as an actual would-be enactor, it SHALL resolve to a U.System or to a role assignment whose holder is a U.System. An episteme, description, publication face, or carrier may participate in the site, but not as the acting bearer.

Episteme non-agency note. If the site is a Description/Episteme, any later enactment still occurs on carriers and/or target systems; the description itself never acts.

#Core construct: ActionInvitationSense

Every in-scope use SHALL resolve to an explicit ActionInvitationSense token.

An ActionInvitationSense token publishes at least:

ActionInvitationSense := ⟨ senseId, siteArity, enactorArity, candidateActionArity, defaultArticulationHint, admissibleArticulationHints, defaultRepresentationSubstrate, admissibleRepresentationSubstrates, defaultNormalForm, admissibleNormalForms, couplingFrameKind, admissibleEvidenceModes, admissibleChangeClasses, bridgePolicy ⟩

Where:

• defaultArticulationHint and admissibleArticulationHints use the current local alias set { open-cue, sketched, option-explicit, hook-explicit }
• defaultRepresentationSubstrate ∈ { ecological-world-coupled, embodied-kinesthetic, latent-distributed, symbolic-local, hybrid }
• admissibleRepresentationSubstrates explicitly declares the lawful publication substrates for the sense;
• defaultNormalForm ∈ { CuePack, ActionOption, OptionSet, PolicyHook }

#A.16 alias-docking note

A.6.A carries articulationHint only as a local alias field.

This field is deliberately not a new formality ladder, not a maturity scale, and not a surrogate for F. Its only job is to preserve local articulation / closure cues until they are docked to A.16 move logic and the explicit C.2.4 / C.2.5 facet owners.

Local articulationHint tokens SHALL dock to A.16 move logic and to the explicit C.2.4 / C.2.5 facet owners one-for-one, and A.6.A SHALL treat them as aliases or publication conveniences only. Until then, local hints SHALL NOT be thresholded, aggregated, or compared across Contexts.

#Normative starter set of sense families

A Context MAY add local senses, but the following starter set is normative as the initial disambiguation menu:

Normative rewrite note.

• In ecological / embodied contexts, bare affords SHALL rewrite to AIS.PhysicalAffordance unless another sense is explicitly declared.
• In interface / alarm / operator-panel contexts, bare action-first phrasing SHALL rewrite to AIS.InterfaceAffordance and/or AIS.ControlOpportunity.
• In epistemic exploration contexts, “this suggests probing / formalizing / reframing” SHALL rewrite to AIS.EpistemicProbe and/or AIS.ClosureAdvance.
• In learned world-model / active-inference / policy contexts, bare “the model wants / the state suggests” SHALL rewrite to AIS.LatentPolicyCue and/or AIS.ControlOpportunity, with the distinction made explicit.
• If the sentence is chiefly about better / worse / fit / merit, use A.6.Q instead of A.6.A.

#Required slots for a conforming actionInvitation

A conforming actionInvitation SHALL make explicit:

1. Site tuple and site-facet docking. What the invitation is about: object, scene, interface object, description, carrier, episode, or control state — with per-member A.7 lane annotations when the tuple is mixed.

2. Invited enactor tuple. Which system / collective system / role-holder is invited to act.

3. Candidate action tuple. What action is being invited.

4. ActionInvitationSense. Which action-oriented family is intended.

5. Coupling frame. The relation-basis under which the invitation is published. Examples: reach envelope, interface state, incident horizon, control horizon, probe pack, open issue set.

6. Detector and/or viewpoint. Who or what detected the cue, and under which viewpoint it is published.

7. Normal form and articulationHint. How the invitation is published and how far it has been articulated.

8. Scope and time when relevant. U.Scope and Γ_time SHALL be explicit when omission changes meaning.

9. Representation substrate when relevant. Especially when comparing ecological, embodied, latent-distributed, and symbolic-local treatments.

10. Witness / evidence mode. Exemplars, sensory traces, probe notes, kinematic data, interface events, controller traces, run logs, or review notes.

#Normal-form discipline

An ActionInvitationSense SHALL declare one lawful default normal form and MAY declare additional admissible normal forms explicitly.

Docking note. Where a published invitation already points to executable artifacts, the record SHOULD reuse existing U.Method / U.MethodDescription / U.WorkPlan identifiers or refs. PolicyHook SHALL always be a hook over pre-existing gate / method / protocol surfaces; it does not mint a new execution, admissibility, or deontic ontology.

ANF-1 — CuePack. Use for early or weakly articulated invitations, especially AIS.PhysicalAffordance, AIS.SocialAffordance, and many cases of AIS.LatentPolicyCue.

A conforming CuePack publishes:

• exemplar or contrast episodes, sensory traces, or probe cues,
• site conditions,
• enactor profile or enactor constraints,
• a small gloss set of candidate actions,
• optional ordinal urgency or salience summaries,
• explicit warning that the cue is not yet a commitment, a selected method, a gate, or work,
• explicit note that witness-bearing does not by itself make the hinted action correct, required, or selected.

ANF-2 — ActionOption. Use when one candidate action tuple is explicit.

A conforming ActionOption publishes:

• one candidate action tuple,
• invited enactor / role,
• local guard sketch,
• expected near-field effect,
• optional U.Method / U.MethodDescription / U.WorkPlan refs when those already exist in-context,
• explicit note that the option is not yet selected, not yet obligatory, and not yet executed.

ANF-3 — OptionSet. Use when several candidate actions coexist.

A conforming OptionSet publishes:

• explicit action members,
• any local comparator, triage rule, or partial order,
• admissible incomparability if no total order is lawful,
• prohibition on hidden scalarisation.

ANF-4 — PolicyHook. Use when the invitation is explicitly bound to an existing controller, gate, playbook, method, or override protocol.

A conforming PolicyHook publishes:

• referenced policy / method / gate / protocol ids (pre-existing owners only),
• applicable guard or trigger conditions,
• ownership / accountable role,
• escalation or override references when relevant,
• explicit note that the hook is a binding surface over existing semantics, not itself a commitment, an admissibility law, or a work occurrence.

#Separation from quality, capability, commitment, and work

A.6.A SHALL prevent the collapse of action invitation language into neighbouring families.

• A statement about better / worse / fit / merit belongs to A.6.Q.
• A statement about what a system can do in general belongs to capability / method description.
• A statement about what must be done belongs to A.6.B (A-* / D-*).
• A statement about what was actually done belongs to A.15 / U.Work.
• If an invitation targets a description/episteme, any later enactment still occurs on symbol carriers and/or target systems; the description itself never acts.
• Mixed sentences that carry both evaluative and invitational content SHALL be split into evaluativeAscription(...) and actionInvitation(...) records, with explicit cross-references when the co-occurrence matters.

Mixed sentences SHALL be split.

Examples:

• “This scene is good for grasping” may require both evaluativeAscription(...) and actionInvitation(...).
• “This alarm requires rollback” is not a lawful final affordance record; it needs explicit gate / duty routing.
• “The robot can grasp this handle” is a capability claim unless the situated site/actor/frame and invitation are made explicit.
• “The operator clicked rollback” is work, not invitation.

#Bridge discipline across traditions

Whenever two traditions are compared using action-first language, the author SHALL publish an explicit bridge stance and loss note.

Allowed bridge stances:

• localRename
• operationalizes
• partialAnalogy
• projection
• nonEquivalent

Examples:

• AIS.PhysicalAffordance - AIS.InterfaceAffordance is usually partialAnalogy, not identity.
• AIS.EpistemicProbe - AIS.ClosureAdvance is usually a progression-by-closure relation, not identity.
• AIS.LatentPolicyCue > AIS.ControlOpportunity is often operationalizes or projection.
• AIS.PhysicalAffordance > PolicyHook in robotics is usually projection under a controller frame.
• Action invitation and quality ascription may co-occur, but co-occurrence is not identity.

#Change lexicon

A conforming pattern SHALL narrate changes with a stable change lexicon aligned to A.6.P:

• declareActionInvitation(...) — create a new explicit action invitation record.
• withdrawActionInvitation(...) — retire a prior record.
• retargetSite(...) — change the site tuple while keeping the same relation family.
• retargetInvitedEnactor(...) — change the invited enactor tuple when that slot is ref-backed.
• reviseAction(...) — change the candidate action tuple by value (or split into the corresponding retargetParticipant(...) form if the local contract makes the action slot ref-backed).
• reviseSense(...) — change the value in the actionInvitationSense slot.
• reArticulate(...) — change the articulationHint while preserving sense family.
• reFrame(...) — change coupling frame.
• reGuard(...) — change guard sketch or hook condition.
• rePolicyHook(...) — change policy / gate / method hook details.
• reView(...) — change detector / viewpoint / view publication.
• rescope(...) — change U.Scope.
• retime(...) — change Γ_time.
• refreshWitnesses(...) — refresh witness bindings.
• changeRelationKind(...) — semantic move to a different relation family; never edit in place silently.

A silent move from invitation to commitment, capability, or work is a breaking semantic change.

A.6.P rewrite note. retargetSite(...) and retargetInvitedEnactor(...) are family-specific refinements of participant retargeting and SHALL be used only when the corresponding slots are ref-backed. reviseAction(...), reviseSense(...), reArticulate(...), reFrame(...), reGuard(...), and rePolicyHook(...) are by-value revisions unless the local contract explicitly declares the corresponding slot as ref-backed, in which case the text SHALL use the matching retargetParticipant(...) form. This preserves A.6.5’s ref-vs-value discipline.

#A.6.B routing template for actionInvitation

When an action invitation becomes boundary-bearing, route it explicitly:

• L — actionInvitation contract skeleton, ActionInvitationSense semantics, normal-form lawfulness, actor/site discipline, bridge stances.
• A — admissibility conditions for using the invitation in selector, triage, automation, or publication lanes.
• D — duties on authors, operators, or owners: lexical firewall, naming the invited actor, naming the hook owner, naming override paths where required.
• E — carrier-anchored witnesses: sensory traces, interface events, probe notes, controller logs, run traces, incident records.

Do not let bare action-first language carry L/A/D/E force by itself.

#Lexical guardrails

In Tech / normative prose:

• bare affords / invites / calls for / actionable / ready for / ripe for / natural next step / the model wants / the interface tells MUST NOT appear without immediate repair;
• actionable insight MUST be rewritten to ActionOption / OptionSet / PolicyHook, or to A.6.Q if the use is primarily evaluative;
• affordance MUST NOT be treated as a monadic property of an object without actor, site, and frame;
• an invitation MUST NOT be presented as if it were already a duty, gate, or work occurrence;
• a latent policy cue MUST NOT be presented as if it were already an explanation;
• articulationHint MUST NOT be read as F, as acceptance status, or as a replacement for A.16 anchors;
• generic Surface facet tokens MUST NOT be introduced inside A.6.A; publication-surface participation must be declared under A.7 / L-SURF, not by widening the A.7 lane set;
• hidden enactor language inside adjectives such as graspable, deployable, actionable, ready SHALL be unpacked;
• quoted metalinguistic uses are allowed, but SHALL be marked as token-under-discussion.

#Progressive elaboration

A.6.A supports monotone elaboration:

1. Start by selecting an ActionInvitationSense and recording rival candidates when ambiguity is live.
2. Declare site, would-be enactor, action, frame, and site-facet docking.
3. Choose a lawful normal form and a local articulationHint when omission would hide articulation state.
4. Add guards, method/policy hooks, and witness bindings.
5. If a CuePack / ActionOption is projected into OptionSet / PolicyHook or docked to A.6.Q / A.6.B / A.15 surfaces, publish an explicit projection / operationalization note rather than silently upgrading the invitation.
6. Add bridges and loss notes if traditions are compared.
7. If the invitation becomes boundary-bearing, emit L/A/D/E routing hooks and, where enactment is implied, route into A.15 surfaces.
8. Never move from invitation into capability, commitment, or work silently.

#Endpoint-first downstream discipline

If a repaired phrase already names a lawful downstream owner such as a gate hook, method surface, work-planning surface, or work occurrence, authors SHOULD publish that owner directly and keep actionInvitation(...) only as the preceding repair record when the invitation semantics themselves still matter. actionInvitation(...) is therefore a post-threshold invitation record, not a shadow substitute for A.6.B, A.15, or gate-owner patterns.

#Archetypal Grounding

#Tell

If a draft says affords, calls for, invites, or actionable, the author has not yet named the action-oriented family.

A conforming post-threshold rewrite publishes one explicit actionInvitation(...) with one ActionInvitationSense, one site tuple, one invited enactor tuple, one candidate action tuple, one coupling frame, one normal form, and explicit articulation / scope / time / substrate qualifiers when they matter. Earlier action-guiding material may still remain outside A.6.A as cue-pack content, a RoutedCueSet, or another typed route-bounded upstream publication until threshold conditions are met.

#Show (System lane)

Draft: “The alarm calls for rollback.”

Repair A — control / incident line

actionInvitation( site = AlarmBundle_AB9 × ServiceState_S7, siteFacetMap = { AlarmBundle_AB9: Carrier, ServiceState_S7: Object }, invitedEnactor = OpsTeam_Phoenix, candidateAction = Enact(MethodDescriptionRef = RollbackRunbook_R41, target = Release_R41), actionInvitationSense = AIS.ControlOpportunity, couplingFrame = IncidentPolicy_IP2 × Horizon_H15m, detector = AnomalyPolicy_AP7, viewpoint = VP.OperationsControl, normalForm = PolicyHook, articulationHint = hook-explicit, scope = U.WorkScope(ProdCluster_EU_1), Γ_time = RunWindow_RW, witnesses = {AlertTrace_91, ErrorBudgetSeries_4} )

Repair B — ecological / robot line

Draft: “This handle affords pulling.”

actionInvitation( site = DoorHandle_17 × DoorState_Closed × ReachEnvelope_RE2, siteFacetMap = { DoorHandle_17: Object, DoorState_Closed: Object, ReachEnvelope_RE2: Description }, invitedEnactor = ServiceRobot_R2, candidateAction = PullAlong(Axis_A1), actionInvitationSense = AIS.PhysicalAffordance, couplingFrame = GripClass_G1 × ClearanceProfile_CP3, detector = PerceptionStack_PS4, normalForm = ActionOption, articulationHint = option-explicit, Γ_time = Window_W1, witnesses = {DepthFrame_883, ContactModelRun_17} )

#Show (Episteme lane)

Draft: “This problem asks for a better question.”

Repair A — epistemic probe line

actionInvitation( site = ProblemFramingEpisode_PF3, siteFacetMap = { ProblemFramingEpisode_PF3: Description }, invitedEnactor = ResearchTeam_A, candidateAction = Enact(MethodDescriptionRef = ContrastiveQuestioning_Q2), actionInvitationSense = AIS.EpistemicProbe, couplingFrame = ExemplarPack_EP3 × OpenIssueSet_O2, detector = Reviewer_A1, normalForm = OptionSet, articulationHint = sketched, representationSubstrate = hybrid, witnesses = {EpisodeNotes_3, CounterexampleCard_2} )

Repair B — closure-advance line

Draft: “The draft is ready for formalization.”

actionInvitation( site = DraftHypothesis_H7, siteFacetMap = { DraftHypothesis_H7: Description }, invitedEnactor = AuthorCollective_C1, candidateAction = Formalize_DS(TypedInvariantSet_V1), actionInvitationSense = AIS.ClosureAdvance, couplingFrame = AmbiguityMemo_8 × ClaimScope_G1, detector = ReviewPanel_R4, normalForm = ActionOption, articulationHint = option-explicit, representationSubstrate = symbolic-local, witnesses = {AmbiguityMemo_8, ReviewCommentSet_5} )

#Bias-Annotation

Lenses tested: Gov, Arch, Onto/Epist, Prag, Did. Scope: Universal for overloaded affordance / action-first language in FPF prose.

• Gov bias: this pattern may tempt authors to smuggle decisions into invitation language. Mitigation: explicit A.6.B routing and obligation barrier.
• Arch bias: this pattern prefers one stable relation family over loose action talk. Mitigation: allow Plain exploratory prose before Tech / normative publication.
• Onto/Epist bias: this pattern insists on separating invitation from evaluation, capability, commitment, and work. Mitigation: explicit bridge stances and mixed-sentence split rules.
• Prag bias: it favors enactor/site/action explicitness, which raises authoring cost. Mitigation: small starter set, normal-form discipline, and copyable rewrites.
• Did bias: repeated rewrites make the pattern teachable, but may over-formalize early cues. Mitigation: CuePack and local articulationHint keep early stages lawful without pretending closure.

#Conformance Checklist (CC-A.6.A)

A text or pattern conforms to A.6.A iff:

1. CC-A.6.A-1 — Explicit post-threshold relation family and explicit sense. Every in-scope post-threshold action-first use resolves to one declared actionInvitation(...) instance and one declared ActionInvitationSense; earlier cue-like material is routed through A.16.1 / B.4.1 instead of being forced into A.6.A prematurely.

2. CC-A.6.A-2 — Explicit site and site-facet docking. The site tuple is explicit; when ambiguous or mixed, the A.7 lane map (Object | Description | Carrier) is explicit.

3. CC-A.6.A-3 — Explicit invited enactor. The invited enactor tuple is explicit.

4. CC-A.6.A-4 — Enactor discipline. When the invited enactor is meant as the actual would-be enactor, it resolves to a U.System or role assignment with system holder.

5. CC-A.6.A-5 — Explicit candidate action. The candidate action tuple is explicit and reviewable.

6. CC-A.6.A-6 — Explicit coupling frame. The coupling frame is explicit.

7. CC-A.6.A-7 — Detector/viewpoint separation. When both matter, detector and viewpoint are not silently collapsed.

8. CC-A.6.A-8 — Lawful normal form. The invitation is published as CuePack, ActionOption, OptionSet, or PolicyHook, with corresponding discipline observed.

9. CC-A.6.A-9 — Articulation-hint discipline. If omission changes meaning, articulationHint is explicit and is not treated as F or as an acceptance state.

10. CC-A.6.A-10 — No invitation-as-obligation. An invitation is not silently published as a duty or gate.

11. CC-A.6.A-11 — No invitation-as-work. An invitation is not silently published as a work occurrence.

12. CC-A.6.A-12 — No capability collapse. A situated invitation is not silently rewritten as a general capability claim.

13. CC-A.6.A-13 — No object-property collapse. Affordance language is not published as a monadic object property when actor/site/frame matter.

14. CC-A.6.A-14 — No hidden scalarisation. OptionSet publication does not introduce a hidden total score or ranking without an explicit comparator / policy.

15. CC-A.6.A-15 — No silent sense rewrite. Sense changes use the declared change lexicon.

16. CC-A.6.A-16 — No silent relation-family switch. Moving from invitation to quality ascription, capability, commitment, or work uses changeRelationKind(...) or an explicit split.

17. CC-A.6.A-17 — Bridge accountability. Cross-tradition parallels publish bridge stance and loss notes.

18. CC-A.6.A-18 — Boundary-routing hook when needed. If the repaired invitation is used for admissibility, commitments, publication, or automation, downstream L/A/D/E hooks are explicit.

19. CC-A.6.A-19 — Lexical firewall. Bare action-first trigger tokens are absent from Tech / normative prose except as quoted metalinguistic discussion.

20. CC-A.6.A-20 — actionInvitation contract skeleton is published. The family-specific RelationKind token resolves to a contract skeleton with SlotSpecs, enactor/site discipline, qualifier expectations, repair paths, witness discipline, admissible change classes, and cross-context policy.

21. CC-A.6.A-21 — Candidate-Set Note is used when ambiguity is live. If the site lane map, enactor lane, relation family, or sense selection is non-obvious, the text records a short Candidate-Set Note before decision-bearing use.

#Common Anti-Patterns and How to Avoid Them

#Consequences

Benefits. This pattern gives FPF a lawful post-threshold repair record family for action-first discourse. It lets embodied, ecological, latent, interface, and control cues be published without pretending they are already commitments, capabilities, metrics, or work.

It also complements A.6.Q cleanly: A.6.Q repairs evaluative ambiguity, while A.6.A repairs action-inviting ambiguity.

Trade-offs / mitigations. The pattern adds authoring overhead and can feel heavy in early exploration.

Mitigation: allow bare action-first language in Plain exploratory notes, but require repair before it enters Tech / normative, boundary, automation, assurance, or publication surfaces.

#Rationale

A.6.A makes one strategic move:

Affordance / action-first language is not treated as a monadic property and not treated as a hidden duty. It is treated as a family of action invitations whose members differ by site, actor, candidate action, coupling frame, substrate, and lawful publication form.

This bridge reading is intentionally neutral: in ecological settings the site is not treated as a literal speaker or norm-giver. “Invitation” is the stable publishable FPF lens for situated opportunity-to-act talk, not a claim that all source traditions use that word or share one ontology.

This gives FPF a lawful path for:

• ecological and embodied affordances,
• interface and operator prompts,
• epistemic “probe this / formalize this / reframe this” moves,
• latent policy cues in learned systems,
• control opportunities in closed loops,

without forcing them into one false universal vocabulary.

It also keeps the larger architecture clean:

• A.6.Q owns evaluative repairs,
• A.6.A owns action-invitation repairs,
• A.6.B owns boundary routing,
• A.15 owns enactment / work,
• A.16 owns articulation / closure progression and lawful moves,
• C.2.3 remains the sole owner of the formality axis F.

#SoTA-Echoing

Recent philosophical and ecological work treats affordances as action-relevant possibilities perceived in engagement and, in some accounts, as invitations for action, rather than as viewpoint-free monadic object properties. A.6.A adopts that relational, action-first stance, adapts it by forcing explicit siteTuple / invitedEnactorTuple / couplingFrame publication, and rejects silent collapse into monadic object labels. (Frontiers, Springer)

Recent empirical review work on affordance perception emphasises attunement and recalibration in person-plus-object systems rather than fixed, context-free labels. A.6.A adopts the need for actor- and situation-specific publication, adapts it into CuePack / ActionOption / OptionSet normal forms, and rejects any assumption that an affordance phrase is already a lawful metric or a universally portable invariant. (Springer)

Current active-inference work frames generative models as supporting action-perception loops and, in many cases, action-oriented models that are for adaptive interaction rather than only detached description. A.6.A adopts the action-oriented emphasis and the separation between model-side cueing and enacted action; it adapts this by making detector and invitedEnactor explicit and by forbidding latent policy cues from counting as work, commitment, or explicit rationale by default. (UCL Discovery)

Current robotics work increasingly uses affordances as intermediate representations between perception/language and low-level action, including compact keypoint or staged affordance plans. A.6.A adopts this as evidence that affordance publication can be a lawful intermediate publication form; it adapts it into ActionOption, OptionSet, and PolicyHook, and rejects silent promotion of such representations into deontic obligation, proof of correctness, or objective value. (Robotics: Science and Systems)

Coverage note. This section already covers the load-bearing relational/action-oriented stance. A fuller canonical corpus package should also bind explicit operator-interaction / operator-alarm / incident-response SoTA-pack material so that operator-facing interface practice is evidenced as directly as the current ecology / active-inference / robotics branch.

#Relations

• Specialises: A.6.P as an RPR pattern for overloaded affordance / action-first language.
• Builds on: A.3/A.7 for enactor discipline and Object≠Description≠Carrier separation; A.15 for keeping invitation distinct from enactment; A.6.B for boundary routing; E.17/E.18 for viewpoint publication.
• Works alongside: A.6.Q for evaluative language; the two are siblings, not substitutes.
• Coordinates with: C.2.2a / A.16 / A.16.1 / A.16.2 / B.4.1 for language-state chart positions, lawful moves before post-threshold repair, and retreat when a published invitation must be reopened; use A.16.0 only when lineage, branch, loss, or handoff history itself must be published as an explicit trajectory account; B.5.2.0 for probe-question cases that are still prompt-shaped; C.2.LS / C.2.4 / C.2.5 / C.2.6 / C.2.7 for language-state facet ownership.
• Must not replace: C.2.3 as the single owner of F.
• Recommends publication via: E.10 / F.17 / F.18 when actionInvitation tokens, starter senses, and red-flag rewrites become shared vocabulary.

#Language-space refactor note

This pattern is scoped to action-invitation repair and routing, not to the whole early cue layer. Early action-guiding material may remain in A.16.1 as cue-pack content, a RoutedCueSet, or another typed route-bounded upstream publication before it stabilizes into actionInvitation(...).

#Canonical downstream seam

actionInvitation(...) should route canonically through A.6.B and A.15 toward gates, commitments, methods, or work. Operator-facing starter senses such as AIS.AlertInterventionCue or AIS.OperatorInterventionCue should not be buried under generic AIS.InterfaceAffordance when human factors and policy hooks materially differ.

#Ownership boundary

Bridge stances, articulation-state owners, and language-state facet axes are referenced by this pattern but remain owned by F.9.1, A.16, C.2.LS, C.2.4, C.2.5, C.2.6, and C.2.7.

#A.6.A:End