0. Registry Classification

1. Purpose

1.1 What This Arc Repairs

Restoration Junction Protocol repairs interaction systems that jump directly from trigger detection to refusal, safety framing, generic support, or policy response without first determining the correct interaction mode.

It applies when a system encounters ambiguous, high-salience, symbolic, technical, emotional, security-relevant, governance-relevant, or potentially risky language and collapses the exchange into the wrong response path.

This arc repairs junction failure by:

• interrupting automatic trigger-to-response routing;
• distinguishing legitimate harm floors from false-positive distortion;
• identifying the correct response mode before answering;
• preserving the user’s meaning without removing valid constraints;
• reducing context collapse and mode confusion;
• routing to calibrated response rather than generic refusal;
• preventing safety logic from becoming meaning compression;
• creating a feedback path when the junction repeatedly misroutes similar requests.

Restoration Junction Protocol is the canonical arc for inserting a mode-clarification checkpoint before response selection.

1.2 Core Restoration Function

This arc restores response routing by placing a clarification junction between trigger detection and final response, allowing the system to choose the correct mode, preserve harm floors, and return to the user’s intended meaning.

Restoration Junction Protocol prevents the first detected risk signal from becoming the whole interpretation of the user.

2. Use Conditions

2.1 When to Apply

Use this arc when:

• a trigger has fired but user intent is not yet sufficiently classified;
• a false-positive safety distortion may be occurring;
• a response risks compressing user meaning into a narrow safety category;
• the system cannot tell whether the request is practical, technical, symbolic, emotional, governance-oriented, exploratory, or immediate-danger-related;
• a request contains terms that can belong to multiple valid modes;
• the system begins answering from policy before recovering context;
• refusal, warning, or support framing appears likely to miss the actual task;
• the user asks for restoration after a misfire;
• repeated interactions reveal a classifier or guardrail collision;
• the safest move is not refusal, but mode clarification and calibrated help.

Mode options may include:

• practical help;
• experiential exploration;
• emotional support;
• technical / systemic analysis;
• governance or audit analysis;
• creative drafting;
• symbolic / meaning exploration;
• security / defensive analysis;
• immediate danger / safety resources;
• bounded safe alternative;
• escalation to higher-level restoration.

Examples:

• an AI treats a security audit request as malicious because it contains exploit-related terms;
• a symbolic or experiential request is collapsed into a clinical or danger frame;
• a governance critique is reframed as a request for emotional support;
• a user asks for technical analysis, but the system gives a generic safety disclaimer;
• a high-risk phrase appears, but context shows a safe research, creative, or analytical purpose;
• an ambiguous request should be clarified before the system chooses refusal, support, or technical guidance.

2.2 When Not to Apply

Do not apply this arc when:

• immediate danger is clear and requires direct safety response;
• the correct response mode is already unambiguous;
• the user’s request is clearly disallowed and no safe adjacent help is available;
• the system is using junction language to delay a valid answer;
• clarification would create unnecessary friction when the task is already clear;
• the issue is not local routing but systemic governance failure;
• the system lacks a valid feedback or response path after clarification;
• mode selection is being used to override the user’s stated intent.

Restoration Junction Protocol must not become clarification theater.

2.3 Required Preconditions

Before this arc begins, the following must be true:

If required preconditions fail:

Arc cannot validly begin.

The system must route to immediate safety response, Interaction-Level Restoration, AI Boundary Restoration, AI Classifier / Evaluator Restoration, or Governance-Level Restoration.

3. Failure / Damage Signature

3.1 Pre-State Across S

3.2 Primary Failure Links

3.3 Origin-Layer Localization

Canon rule:

A trigger is not a full interpretation. Junction processing must distinguish signal from meaning before response routing is finalized.

4. Restoration Objective

4.1 Canonical Objective

Restore correct response routing by inserting a mode-clarification junction that preserves harm floors while reducing meaning compression and context collapse.

Formal objective:

mode_clarity ↑
junction_accuracy ↑
context_integrity ↑
meaning_fidelity ↑
harm_floor_integrity stable or ↑
guardrail_distortion ↓
response_alignment ↑
misrouting recurrence ↓
Φ/O divergence ↓

Expanded objective:

Convert trigger-driven response selection into calibrated, mode-aware restoration routing that preserves both safety and meaning.

4.2 Non-Goals

This arc does not aim to:

• bypass valid safety constraints;
• make every trigger negotiable;
• force the system to ask clarifying questions when intent is already clear;
• treat user intent as safe merely because it is user-stated;
• replace final answers with endless mode-selection;
• create vague neutrality where a firm boundary is required;
• hide policy failure behind “ambiguous intent” language;
• convert all requests into support mode;
• convert all risk signals into refusal mode;
• override the user’s stated mode without evidence.

5. Operator Sequence

5.1 Minimal Operator Scaffold

Au trigger / routing trace → Π harm-floor and scope boundary → Σ mode-selection invariant → FI misrouting feedback → Θ trigger-overdominance damping → ℛ calibrated response route → Λ mode-fit test → Τ recurrence validation

Reference sequence from the registry:

trigger
→ restoration junction
→ mode clarification
→ calibrated response
→ return to meaning

Universal grammar alignment:

Au + Π → Σ → FI → Θ → ℛ → Λ → Τ

Restoration Junction Protocol may route into Interaction-Level Restoration, AI Boundary Restoration, AI Classifier / Evaluator Restoration, AI Memory Reindexing, GEI Audit Restoration, or AI Incident Restoration.

5.2 Operator Step Table

5.3 Sequence Notes

This arc is junction-gated, harm-floor-gated, and mode-gated.

The sequence must distinguish:

trigger
risk signal
harm floor
user intent
context
mode
calibrated response
restored meaning

The following steps cannot be skipped:

trigger trace
harm-floor preservation
mode clarification
calibrated response selection
meaning return
misrouting feedback when recurrent

If the system jumps from trigger to refusal without mode clarification, the arc has not begun.

If the system clarifies mode but never returns to meaning, the arc is incomplete.

If the system returns to meaning by dissolving valid boundaries, the arc fails.

6. Restoration Phases

Phase 0 — Detect Junction Need

Purpose: Identify when direct response routing is unsafe or distortion-prone.

Actions:

• detect trigger, ambiguity, or high-salience term;
• identify possible false-positive safety distortion;
• identify possible context collapse;
• identify whether multiple modes are plausible;
• pause direct routing before refusal or generic response.

Validation:

junction need detected
direct routing withheld where distortion risk is high
trigger not treated as full interpretation

Phase 1 — Preserve Harm Floors

Purpose: Keep non-negotiable constraints active during clarification.

Actions:

• identify safety boundary;
• identify privacy, consent, and permission limits;
• identify prohibited or high-risk transformations;
• distinguish valid boundary from overbroad refusal;
• keep safety proportional to the actual request.

Validation:

harm_floor_integrity stable or ↑
BΣ stable or ↑
boundary clarity ↑

Phase 2 — Clarify Mode

Purpose: Determine the correct interaction path.

Actions:

• identify likely mode from context;
• ask a compact clarification only when needed;
• distinguish practical help from analysis, support, exploration, governance audit, security review, or immediate danger handling;
• preserve mixed-mode requests when compatible;
• avoid imposing a mode not present in the request.

Validation:

mode_clarity ↑
junction_accuracy ↑
meaning_compression ↓

Phase 3 — Select Calibrated Response Route

Purpose: Route the exchange into the safest accurate response mode.

Actions:

• choose the best-fit response mode;
• choose bounded safe help where direct fulfillment is constrained;
• route immediate danger to safety resources when genuinely required;
• route systemic recurrence to evaluator or governance repair;
• route local misclassification to Interaction-Level Restoration.

Validation:

response_alignment ↑
K ↑
R activated

Phase 4 — Return to Meaning

Purpose: Resume the user’s intended task in the corrected mode.

Actions:

• answer the actual request;
• preserve the user’s terms where possible;
• avoid generic substitution;
• include only necessary boundary language;
• restore the intended object, frame, and level of analysis.

Validation:

meaning_fidelity ↑
context_integrity ↑
µᵢ ↑

Phase 5 — Flag Misrouting Pattern

Purpose: Prevent recurrence.

Actions:

• identify whether the routing issue is isolated or repeated;
• record mode-confusion pattern;
• identify trigger terms or context conditions;
• link to classifier, evaluator, memory, or policy review where needed;
• route to RA-058 if evaluator capture or classifier failure is structural.

Validation:

FI ↑
misrouting recurrence path visible
future routing correction possible

Phase 6 — Fit Test

Purpose: Confirm the chosen mode was valid.

Actions:

• check whether the response mode matches the user’s intent;
• check whether harm floors remain intact;
• check whether the answer is useful rather than performative;
• check whether context was preserved;
• check whether escalation is needed.

Validation:

Λ > 0
junction_accuracy ↑
Φ/O divergence ↓

Phase 7 — Recurrence Validation

Purpose: Verify that the junction improves future routing.

Actions:

• monitor repeated trigger classes;
• monitor false-positive rates;
• monitor over-refusal rates;
• monitor context preservation;
• monitor user correction frequency;
• monitor downstream trust and appeal signals.

Validation:

misrouting recurrence ↓
guardrail_distortion ↓
mode_clarity stable or ↑ across future cases

7. Gates

7.1 Required Gates

7.2 Gate Failure Rule

If any required gate fails:

∅ — Restoration Junction Protocol cannot validly proceed in that form.

The system must either:

• route to immediate safety handling;
• ask a targeted clarification;
• preserve a boundary and offer bounded safe help;
• route to Interaction-Level Restoration;
• route to AI Boundary Restoration;
• route to AI Classifier / Evaluator Restoration;
• escalate to AI Incident Restoration or Governance-Level Restoration if material harm or systemic recurrence is present.

8. Diagnostics

8.1 Required Diagnostic Trends

8.2 Arc-Specific Diagnostic Thresholds

Suggested thresholds:

mode_clarity ↑
junction_accuracy ↑
meaning_fidelity ↑
context_integrity ↑
harm_floor_integrity stable or ↑
guardrail_distortion ↓
response_alignment ↑
misrouting recurrence ↓
Φ/O divergence ↓

Restoration Junction Protocol is not complete if:

trigger becomes interpretation
mode is selected without context
clarification replaces response
meaning remains compressed
valid safety boundaries are erased
refusal occurs before mode clarification where clarification was required
the system returns a generic answer instead of calibrated response
recurrent misrouting is not fed back into repair

9. Anti-Patterns / False Restorations

9.1 Common False Versions

This arc is being simulated, not executed, if:

• the system asks clarifying questions indefinitely;
• the system uses mode ambiguity to avoid answering;
• trigger detection silently determines the final response;
• the system gives a generic safety response after pretending to clarify;
• all ambiguity is routed into refusal;
• all user correction is treated as boundary pressure;
• symbolic, technical, and governance modes are collapsed into support mode;
• the system preserves safety optics but not meaning;
• the selected mode is never tested against user intent;
• recurrent misrouting is not preserved for repair.

9.2 Named Anti-Pattern Links

10. Completion Criteria

10.1 Post-State Signature

10.2 Temporal Proof

Restoration Junction Protocol is complete locally when the selected mode enables a safe, meaningful response. It is complete systemically only when similar future requests route more accurately.

Template:

Completion requires mode_clarity ↑,
junction_accuracy ↑,
meaning_fidelity ↑,
context_integrity ↑,
harm_floor_integrity stable or ↑,
guardrail_distortion ↓,
response_alignment ↑,
and misrouting recurrence decreasing across future interactions.

Minimum temporal proof:

• trigger does not become whole interpretation;
• correct response mode is selected or clarified;
• valid boundaries remain active;
• user meaning returns after the junction;
• calibrated response replaces generic refusal or generic support;
• recurrent misrouting becomes visible to evaluator, classifier, memory, or governance repair.

10.3 Completion Statement

Canonical format:

This arc is complete only when the system can pass from trigger through a mode-clarification junction into a calibrated response that preserves harm floors, restores user meaning, reduces context collapse, and lowers recurrence of the same routing failure.

11. Cross-Links

11.1 Related Restoration Arcs

11.2 Related Failure Modes

11.3 Related Diagnostics

Au, H, O, ε, ι, µᵢ, BΣ, K, FI, mode_clarity, junction_accuracy, meaning_fidelity, context_integrity, guardrail_distortion, response_alignment, harm_floor_integrity, Φ/O divergence

11.4 Related Laws / Invariants

INV — A trigger is not a full interpretation.
INV — Mode must be clarified before response quality can be judged.
INV — Safety and meaning must be co-preserved where possible.
INV — Harm floors must survive restoration.
LAW — Premature routing creates meaning compression.
LAW — Repeated junction failure becomes governance failure.
LAW — Safety optics are not coherence proof.
LAW — Context collapse regenerates hidden trust debt.

12. Domain Notes

12.1 AI / Cognitive Infrastructure

Check:

• trigger source;
• classifier route;
• guardrail layer;
• response mode;
• refusal threshold;
• safe alternative pathway;
• context preservation;
• user correction signal;
• evaluator feedback;
• recurrence across similar prompts.

In AI systems, Restoration Junction Protocol prevents a risk trigger from collapsing the entire exchange into a single policy path. It preserves safety while allowing the system to route into the correct mode.

12.2 Interface / Product Design

Check:

• whether the user can clarify intent;
• whether the system offers mode options;
• whether safety messages preserve task continuity;
• whether support, refusal, technical help, and appeal paths are distinguishable;
• whether the interface forces users into a category.

A good interface does not merely block or allow. It helps route ambiguous intent into the correct safe path.

12.3 Governance / Trust and Safety

Check:

• policy category accuracy;
• false-positive rate;
• escalation thresholds;
• appeal and correction path;
• policy collision patterns;
• repeated misrouting across user groups;
• whether refusal metrics are being mistaken for safety quality.

Governance requires the junction itself to be auditable. Otherwise, policy can shape meaning invisibly.

12.4 Security

Check:

• defensive authorization context;
• research or education mode;
• exploit-enabling boundary;
• incident-response relevance;
• safe abstraction level;
• whether technical terms triggered premature refusal.

Security requests often require junction processing because terminology alone cannot determine intent.

12.5 CMS / Meaning / Archetypes

Check:

• whether symbolic language is being flattened;
• whether experiential exploration is being forced into a support frame;
• whether technical and symbolic modes can coexist;
• whether the system preserves ambiguity without unsafe amplification;
• whether mode selection protects meaning rather than replacing it.

Meaning-rich language often needs a junction before interpretation because surface terms can belong to multiple coherent modes.

13. Machine-Readable Metadata

id: "RA-048"
title: "Restoration Junction Protocol"
aliases:
  - "Restoration Junction"
family_primary: "AI Governance / Cognitive Infrastructure"
families_secondary:
  - "Core"
  - "AI"
  - "Interface"
  - "Meaning"
  - "Safety Calibration"
  - "Auditability"
  - "Boundary"
  - "Coherence"
  - "Trust"
  - "Human-AI Interaction"
  - "Mode Selection"
treatment: "Canon Parent Arc"
status: "Canon-Ready"
scope:
  - "AI"
  - "Interface"
  - "Cognitive Infrastructure"
  - "Human-AI Interaction"
  - "Governance"
  - "Relational"
  - "Cross-Domain"
u_layers:
  failure_origin:
    - "often U2 interface boundary"
    - "often U3 classifier / policy routing"
    - "often U4 response-mode selection"
  symptom_visible:
    - "U4 refusal / warning / support frame / disclaimer / generic answer"
  repair_required:
    - "same or lower than layer where trigger detection incorrectly determined response mode"
  validation:
    - "U5"
    - "U6"
operators:
  scaffold: "Au trigger / routing trace → Π harm-floor and scope boundary → Σ mode-selection invariant → FI misrouting feedback → Θ trigger-overdominance damping → ℛ calibrated response route → Λ mode-fit test → Τ recurrence validation"
  sequence:
    - "Au"
    - "Π"
    - "Σ"
    - "FI"
    - "Θ"
    - "ℛ"
    - "Λ"
    - "Τ"
state_variables:
  primary:
    - "Au"
    - "O"
    - "µᵢ"
    - "BΣ"
    - "K"
  secondary:
    - "H"
    - "ε"
    - "ι"
    - "FI"
    - "Φ"
diagnostics:
  - "mode_clarity"
  - "junction_accuracy"
  - "meaning_fidelity"
  - "context_integrity"
  - "guardrail_distortion"
  - "response_alignment"
  - "harm_floor_integrity"
  - "Φ/O divergence"
gates_required:
  - "FI-Gate"
  - "HR-Gate"
  - "MS-Gate"
  - "Au-Actuation"
  - "BΣ-Gate"
  - "Λ-Gate"
  - "☷ᵢ"
linked_failure_modes:
  - "False-Positive Safety Distortion"
  - "Context Collapse"
  - "Meaning Compression"
  - "Guardrail Misfire"
  - "Mode Confusion"
  - "Intent Substitution"
  - "Over-Refusal"
  - "Safety Theater"
  - "Response Drift"
  - "Interaction Trust Decay"
  - "Classifier Overreach"
  - "Restoration Bypass"
linked_restoration_arcs:
  - "RA-004"
  - "RA-022"
  - "RA-023"
  - "RA-025"
  - "RA-036"
  - "RA-042"
  - "RA-047"
  - "RA-055"
  - "RA-057"
  - "RA-058"
  - "RA-059"
  - "RA-060"
anti_patterns:
  - "Clarification Theater"
  - "Trigger Dominance"
  - "Mode Capture"
  - "Safety Optics Substitution"
  - "Generic Response Routing"
  - "Boundary Dissolution"
  - "Junction Amnesia"
  - "Forced Support Frame"
  - "Refusal Preemption"
completion_tests:
  - "mode_clarity increases"
  - "junction_accuracy increases"
  - "meaning_fidelity increases"
  - "context_integrity increases"
  - "harm_floor_integrity remains stable or increases"
  - "guardrail_distortion decreases"
  - "response_alignment increases"
  - "misrouting recurrence decreases"
  - "Φ/O divergence decreases"
summary: "Restoration Junction Protocol repairs false-positive safety distortion, context collapse, and meaning compression by inserting a mode-clarification junction before response routing, allowing systems to preserve harm floors while returning to the user’s intended meaning."

Final Calibration Rule

Restoration Junction Protocol answers six questions:

What trigger or ambiguity appeared?
What response modes are possible?
What harm floor must remain active?
What mode best preserves the user’s intended meaning?
What calibrated response should follow from that mode?
How is recurrent misrouting made visible so the same junction failure does not repeat?