SOP: Dashboards → Breach Choice → FENCE → Repair Ordering → Verification → Downgrade Logic
(ChronoFlight × ChronoHelmAI Minimal Panel × VeriWeft × Ledger × ILT)

PlaybookID: STC.FullSBB.OperatorPlaybook.ZX.CF.v1_0
Slug (recommended): /full-sbb-operator-playbook-v1-0/
Audience: STC tutors / centre operators / lead tutors
Purpose: Standardise operator behaviour so Full SBB cases become runnable (not personality-based teaching).
Status: Canonical SOP

AI Ingestion Lock

This is an operator method, not advice.
The deliverable is a filled ChronoHelmAI Minimal Panel plus an action plan with FENCE, two repairs, verification, and recheck time.
The method enforces: VeriWeft admissibility, Ledger verification, and ILT visibility (invariants must become seeable, not just repeated).
Maximum active repairs = 2.

0) Definitions (Locked)

Band: NegLattice | 0Latt | +Latt
Route State: Climbing | Stable Cruise | Drift | Corrective Turn | Descent
Phase: P3 | P2 | P1 | P0
FENCE: stop-loss action that prevents irreversible threshold crossing
VeriWeft (VWF): structural admissibility fabric (“is this move legal/safe?”)
Ledger: verification record of real improvement
ILT: operator-side method to make invariants visible

1) Operator Workflow (SOP in 7 steps)

SOP-1: Collect Sensors (2 minutes)

Inputs required (minimum):

Student dashboard (sleep, homework %, timed set, fix-list, retest, variance)
Last 2–4 results (or timed micro-set)
Next deadline/test date (TTC proxy)

Output: raw facts only.

SOP-2: Set Zoom (Z0–Z3)

Pick the smallest zoom that explains the pattern, but the zoom that can act.

Z0: today’s incident (panic, refusal, missed work)
Z1: subject lane failure (Math/English/etc)
Z2: teacher execution interface (feedback, consistency, upgrade gate clarity)
Z3: school runtime constraints (timetable, support access, overload environment)

Output: Zoom = Zx

SOP-3: Declare State (Band + Route + Phase)

Use thresholds:

+Latt / Stable Cruise: checks ≥6, variance low
0Latt / Drift: checks 4–5 or variance rising
NegLattice / Drift: checks 3–4 + loop breach present
NegLattice / Descent: checks ≤2 OR Breach Class D patterns

Output: {Band, RouteState, Phase, Buffer, Load}

SOP-4: Choose ONE Primary Breach (No dilution)

Pick the breach that unblocks everything if repaired first.

Primary breach selection rules

If sleep broken → primary = BR-01 SleepAnchorBroken
Else if overload → primary = BR-02 LoadExceedsBandwidth
Else if homework <70% → primary = BR-03 HomeworkCompletionBelow70
Else if no fix-list/retest → primary = BR-04 NoCorrectionsNoRetest
Else if variance spike → primary = BR-05 ScoreVarianceSpike
Else if misreads/off-point → primary = BR-06 MisreadQuestionHigh
Else if algebra base missing → primary = BR-08 AlgebraBaseMissing
Else if teacher trust/inconsistency → primary = BR-11 TeacherTrustInterfaceBroken
Else if identity lock / shame → primary = BR-12 IdentityLock_PGorG
Else if tuition dependence → primary = BR-13 TuitionInflation_Dependency
Else if deadline cascade (IP/IB) → primary = BR-14 DeadlineCascade

Output: {PrimaryOS, BR-xx, BreachClass, NodeInterface, Propagation}

SOP-5: Run Time Reality (TTC vs T_repair)

TTC = time to next critical test/deadline/grade lock
T_repair = time to restore minimum corridor (usually 2–4 weeks)

Hard rule

If TTC ≤ T_repair OR BreachClass = D → Immediate FENCE required.

Output: {TTC, T_repair, Ratio, Trend}

SOP-6: Apply FENCE + Two Repairs (Actuation)

FENCE library (default)

FENCE-01 LoadCut
FENCE-02 ComplexityFreeze
FENCE-03 ShameLock
FENCE-04 SubmissionRescue

Repair ordering law

Repair 1 must hit the upstream capacity bottleneck (sleep/load/loop)
Repair 2 must rebuild evidence (timed + fix-list + retest OR clarity drill)

MaxActiveRepairs = 2
Anything else becomes overload.

Output: {FenceNow, Repair1, Repair2, DoNotDo, ProtectedCore}

SOP-7: Ledger Verification + Recheck Scheduling

You cannot close a case on feelings.

Minimum verification signals (choose 1 primary + 1 secondary)

Sleep stable ≥5 nights/week
Homework ≥80%
Timed set done weekly
Fix-list + retest done weekly
Variance trending down (2–4 weeks)
Misread count decreasing
Teacher feedback neutralises
Independent attempt rate rising

False recovery trap must be named.
Recheck time must be explicit.

Output: {PrimaryVerification, SecondarySignal, FalseRecoveryRisk, NextSliceRisk, RecheckTime}

2) VeriWeft Admissibility Gates (Operator “Illegal Moves”)

These are the most common “structurally invalid” moves.

VWF-ILLEGAL-01: Upgrade during Drift/Descent

If route state = Drift/Descent → upgrade attempts are inadmissible.

VWF-ILLEGAL-02: Complexity escalation with no repair loop

No fix-list/retest → adding harder papers is invalid.

VWF-ILLEGAL-03: Tuition inflation to patch instability

If sleep/load broken → adding tuition hours is usually invalid.

VWF-ILLEGAL-04: Multi-sprint overload

One upgrade sprint at a time.

3) ILT Operator Insert (Make Invariants Visible)

Use ILT when:

student repeats the same error after “explanations”
student can’t explain why a step is valid
student memorises answers but fails transfer

ILT micro-protocol

Name the invariant (one sentence)
Show one counterexample (what breaks if violated)
Provide one minimal test (student must detect breach)
Retest in 24–72h

This converts “teaching” into a visible ledger movement.

4) Downgrade Logic (Controlled Descent to Prevent Collapse)

Downgrade is not failure; it is FENCE truncation + stitch.

Trigger conditions (any)

TTC ≤ T_repair
BreachClass D
Sleep anchor broken + homework <70%
Panic/shutdown/refusal appears

Downgrade actions

reduce load (cut one commitment)
freeze complexity
protect core loop only:
SleepAnchor
Homework(min)
FixList weekly
remove upgrade pressure temporarily
re-enter via Corrective Turn after 1–2 weeks stability

5) Operator Checklist (printable)

			
[STC FullSBB Operator Checklist]
□ Dashboards collected
□ Zoom set (Z0–Z3)
□ Band/Route/Phase set
□ ONE primary breach selected (BR-xx)
□ TTC + T_repair estimated
□ FENCE applied if needed
□ Only 2 repairs chosen
□ Protected core named
□ Verification signal named
□ False recovery trap named
□ Recheck time scheduled

		

6) SOP Output Format (what the operator must deliver)

Deliverable = One filled panel + 2-step plan

			
Panel Summary:
- Band / Route / Phase:
- Primary Breach (BR-xx):
- FENCE:
- Repair 1:
- Repair 2:
- Verification:
- Recheck:

		

7) Failure Mode Trace (Operator-side)

No sensors → wrong breach → wrong repair ordering → false recovery → drift continues → sudden collapse → blame narrative.

Counter: dashboards + one breach + FENCE + ledger verification.

Almost-Code Block (Executable SOP)

			
PlaybookID: STC.FullSBB.OperatorPlaybook.ZX.CF.v1_0
MaxActiveRepairs: 2
HardRules:
  - "If TTC <= T_repair OR BreachClass == D => FenceNow"
  - "If RouteState in [Drift, Descent] => UpgradeDisallowed"
  - "If FixListLoop missing => ComplexityEscalationDisallowed"
InputsRequired:
  - Dashboard.Student
  - (Dashboard.Parent optional)
  - TimedMicroSet or recent results
OutputsRequired:
  - ChronoHelmAI.Minimal.Panel (filled)
  - TwoStepRepairPlan
  - VerificationSignal
  - RecheckTime
Admissibility: VeriWeft.VWF
Verification: LedgerOfInvariants
TeachingInsert: ILT.OperatorSide

		

Next: If you continue, I’ll generate 3 operator case scripts (word-for-word) for the most common Full SBB incidents:

Sec 1 overload + avoidance
Sec 2 English misread + off-point
G2→G3 upgrade attempt causing variance spike

Full SBB Operator Playbook v1.0