---
## The One-Panel Engine (how the whole site runs)
Copy block:

---
## Step 1 — Diagnose (do this first)
Go here: **/btt-diagnosis-backtest-engine-v1-0/**
You’ll run either:
* **15-minute triage** (fast leak detection), or
* **45–60 minute deep backtest** (full leak map + baselines)
**Output you should get:**
* LeakMap (Z0–Z3)
* Sensor baselines (accuracy, speed, transfer, load stability)
* Recommended starting window (Freeze / Sandbox / Deploy)
---
## Step 2 — Repair (ERCO chooses what to fix and fences it)
Go here: **/btt-erco-repair-overlay-v1-0/**
ERCO produces a single output that matters:
* **TruncationActions** (stop the spiral)
* **StitchingActions** (repair fundamentals + rails)
* **RetestSchedule** (prove repair)
* **WindowPlan** (Freeze / Sandbox / Deploy)
This is the difference between “practice more” and **repairing the actual leak**.
---
## Step 3 — Schedule (ChronoHelm turns the plan into 12 weeks)
Go here: **/btt-chronohelmai-scheduler-v1-0/**
ChronoHelm gives:
* weekly plan
* daily minimum
* load ramp
* stop-loss rules (auto revert if instability returns)
---
## Step 4 — AVOO Academy Ground Zero (role training that prevents chaos)
Go here: **/avoo-ground-zero-v1-0/**
This is the Ground Zero platform:
**where student capability is trained as roles**:
* **Operator**: stable execution under time
* **Architect**: controlled novelty and transfer
* **Oracle**: predict and prevent mistakes
* **Visionary**: set targets under constraints
### The safety rule (non-negotiable)
If a student is unstable (P0/P1), too much “creative choice” will collapse execution.
So we enforce **Choice Budget** and **Windows**.
Copy block:

---
## Step 5 — Pick your Subject Engine (English / Math / Science)
Choose the lane you need most right now:
* **English Control Tower** → **/english-control-tower-v1-0/**
* **Math Control Tower** → **/math-control-tower-v1-0/**
* **Science Control Tower** → **/science-control-tower-v1-0/**
Each subject engine is AVOO-wired:
* Operator rails first
* Architect corridors only when fenced and stable
* Retests prove upgrades before difficulty increases
---
## Step 6 — Parents: minimum rails (help without becoming a tutor)
Go here: **/parents-minimum-rails-v1-0/**
This gives a low-effort system:
* 10-minute daily check
* “show me your plan” prompt
* weekly retest day
* protect sleep and routines (capacity)
---
## Step 7 — Global Projection (Bukit Timah → World)
Go here: **/btt-global-projection-btm-to-world-v1-0/**
This exports:
* the Diagnose/Repair/Retest protocol
* sensor dictionaries
* AVOO pack schemas
* Z5 country templates (adapters, not claims)
---
## Quick Routing: If you don’t know where to start
### If performance collapsed suddenly (panic, avoidance, freeze)
Start:
1. Diagnose
2. ERCO
3. ChronoHelm
4. Operator packs only (Freeze window)
### If student is “okay at homework but fails mixed papers”
Start:
1. Diagnose (deep backtest)
2. ERCO
3. Math/English control tower
4. Reopen Architect tasks only when stable
### If student “has correct ideas but low marks”
Start:
1. Diagnose
2. ERCO
3. Science Operator pack (frames + keywords)
---
## Canonical Runtime Footer (paste onto major pages)

ID: DEFLOCK_BTT_RUNTIME_01
Version: v1.2
BTT Runtime:
BukitTimahTutor.com is a running Education hub that converts:
Diagnose → Repair → Retest → Stabilise (P0→P3),
and exports its protocols as templates (Z5).
EducationOS Core Law (Rate Dominance):
Learning improves when Ġ_repair > Ḋ_decay under real load.
AVOO:
Role lattice for capability building:
Architect (generate) → Visionary (aim) → Oracle (predict) → Operator (execute)
Ground Zero:
INST:SGP:BTM:BTT is the canonical testbed where AVOO packs are installed,
measured under load, fenced for safety, then exported as templates.

ID: DIAG_BTT_ONE_PANEL_01
INPUTS
  StudentState + Goal + TimeBudget
  BacktestResults (accuracy/speed/transfer/load)
  SensorDict (EDU + AVOO)
PROCESS (Closed Loop)
  1) Diagnose (Backtest → leak map)
  2) ERCO decides (Truncate/Stitch + Windows)
  3) ChronoHelm schedules (weekly plan + ramps)
  4) Execute (Operator rails + bounded Architect sandbox)
  5) Retest (same sensors, same thresholds)
  6) Promote (difficulty + transfer) OR revert (truncate/stitch)
OUTPUTS
  RepairPlan + RetestSchedule + WindowPlan
  Subject Engines (ENG/MATH/SCI)
  AVOO Academy Packs (A/V/O/O)
  Projection Templates (BTM → World)

ID: INSTALL_BTT_01
Version: v1.2
InstallOrder:
1) BTT Control Tower (this page)
2) Diagnosis / Backtest Engine
3) ERCO Repair Overlay (decide + enforce windows)
4) ChronoHelmAI Scheduler (sequence + guardrails)
5) AVOO Academy (Ground Zero packs + choice budget)
6) Subject Engines:
   - English Engine (Composition + Language)
   - Math Engine (Primary/Secondary/A-Math)
   - Science Engine
7) Parent Toolkit (minimum-effort support rails)
8) Global Projection Templates (Z5 shells)

ID: BTT_ID_PACK_01
PLACE:SGP:BTM
INST:SGP:BTM:BTT
LANE:EDU
LANE:VOCAB
LANE:MATH
LANE:SCI
LANE:MIND
LANE:META

ID: SENSOR_DICT_BTT_01
Version: v1.0
S1  AccuracyRate (by topic + question-type)
S2  TimePerMark (speed stability)
S3  ErrorTaxonomy (careless / concept / misread / method / language)
S4  RetentionHalfLife (days-to-decay)
S5  TransferScore (unseen variants)
S6  LoadStabilityIndex (timed variance)
S7  PanicSpikeIndex (freeze/avoidance)
S8  PlanningCompleteness (before writing/solving)
S9  RevisionDepth (quality delta after rewrite)
S10 HelpSeekingLatency (delay-to-fix)
Thresholds (default bands; tune by level):
T1 AccuracyFloor = 0.70
T2 TransferFloor = 0.60
T3 LoadFloor = stable timed performance (variance capped)
T4 TimePerMark = within target band

ID: SENSOR_DICT_AVOO_01
Version: v1.0
A1 RoleMix (A/V/O/O proportions)
A2 ChoicePressureRatio ρ = S_inj / S_cap
A3 NoveltyBandwidth (distance from standard forms)
A4 FascinationPullIndex (curiosity pull)
A5 CorridorRetention (48h retention of new corridor)
A6 OperatorSlipRate (error inflation after novelty)
A7 PhaseShearProxy (variance under timing)
Thresholds:
T_A2  ρ_max_deploy = 1.0
T_A6  Slip_max = +15% vs baseline
T_A7  Shear_max = level-tuned cap

ID: FENCE_BTT_01
Version: v1.0
Fence Rule:
If (LoadStabilityIndex < T3) OR (ρ > T_A2) OR (Shear > T_A7):
  → TRUNCATE novelty/difficulty
  → STITCH fundamentals + rails
  → RETEST
  → Resume ramp only after stability returns
Windows:
Freeze Window (Operator Stability):
  - zero novelty
  - accuracy + speed rails
  - timed stability
Sandbox Window (Architect Exploration):
  - bounded novelty tasks only
  - short cycles, reversible
  - allowed only if OperatorSlip <= T_A6
Deploy Window (Exam Execution):
  - proven patterns only
  - timed sets + checklists
  - disallow untested novelty

ID: BTT_MODULE_CONTROL_TOWER_01
Version: v1.2
Contract:
- Provide install order + routing to all engines.
- Keep one canonical sensor dictionary.
- Provide “run loop” instructions for humans + LLMs.
Inputs:
- StudentProfile
- Goal + Timeline
- BacktestResults
- SENSOR_DICT_BTT_01 + SENSOR_DICT_AVOO_01
Outputs:
- RepairPlanRouter
- WindowPlanRouter
- EngineRouter (ENG/MATH/SCI + AVOO packs)
- ProjectionIndex (BTM → World templates)

ID: BTT_MODULE_BACKTEST_01
Version: v1.0
Contract:
Convert confusion into a measurable leak map + next actions.
Modes:
- TriageScan (15 min)
- DeepBacktest (45–60 min)
Outputs:
- LeakMap:
  Z0 Mistake patterns
  Z1 Topic weaknesses
  Z2 Habit/routine gaps
  Z3 Load failure triggers
- Baselines for sensors S1–S10
- Recommended WindowPlan entry state (Freeze/Sandbox/Deploy)

ID: BTT_MODULE_ERCO_01
Version: v1.0
Contract:
Choose actions using sensors + thresholds and enforce windows.
Inputs:
- LeakMap
- SensorReadings
- TimeBudget
Outputs:
- RepairPlan:
  - TruncationActions
  - StitchingActions
  - RetestSchedule
- WindowPlan:
  - Freeze / Sandbox / Deploy durations
- EscalationRules (when to promote difficulty)

ID: BTT_MODULE_CHRONOHELM_01
Version: v1.0
Contract:
Sequence RepairPlan into weeks/days with guardrails.
Outputs:
- WeeklyPlan (12-week default)
- DailyMinimum (non-negotiable)
- LoadRamp (week 1 → week 12)
- StopLoss rules:
  if 2 consecutive bad cycles → auto revert to Freeze + Stitch

ID: BTT_MODULE_AVOO_ACADEMY_01
Version: v1.0
Anchor:
  PLACE:SGP:BTM
  INST:SGP:BTM:BTT
Contract:
Train AVOO roles as executable skills under load,
fenced by Choice Budget so Operator stability is never sacrificed.
Inputs:
- StudentProfile (Level, Phase, Timeline)
- Domain (ENG/MATH/SCI/VOCAB/MIND/META)
- SensorDict EDU + AVOO
- Backtest baselines
Outputs:
- RoleMixPlan (A/V/O/O this week)
- ChoiceBudget (ρ constraints)
- TaskDeck (drills + prompts + timed sets)
- WindowPlan (Freeze/Sandbox/Deploy)
- ExportPack template (Z5)

ID: AVOO_RULE_CHOICE_BUDGET_01
Version: v1.0
Definitions:
S_inj = novelty/choices injected per unit time
S_cap = capacity (attention + working memory + routines)
ρ = S_inj / S_cap
Rule:
If (P ≤ P1) OR (LoadStabilityIndex < T3):
  - Architect tasks only in Sandbox Window
  - Operator dominates Deploy Window
Else (P2–P3 stable):
  - Expand Architect tasks gradually
  - Promote only if OperatorSlip and Shear remain below thresholds

ID: AVOO_PACK_REGISTRY_01
Version: v1.0
PACKS (Ground Zero):
PACK_A_ENG_01  Architect of Language (bounded novelty corridors)
PACK_O_ENG_01  Operator English Execution (timed writing + checklists)
PACK_A_MATH_01 Architect of Math (variation ladder, invariant-preserving)
PACK_O_MATH_01 Operator Math Execution (method rails + speed rails)
PACK_A_SCI_01  Architect of Science (explanation corridors, hypothesis play)
PACK_O_SCI_01  Operator Science Execution (answer frames under time)
PACK_META_01   Universal Architect Pack (transfer across domains)
PACK_META_02   Oracle Pack (prediction + error forecasting)
PACK_META_03   Visionary Pack (goal + constraint targeting)

PACK_SCHEMA_v1:
- Definition
- Mechanism
- TaskDeck (Z0 drills / Z1 routines / Z2 cadence)
- Sensors + thresholds
- Failure trace
- Repair loop (truncate/stitch/retest)
- Export template (Z5)

ID: EDU_ENGINE_ENG_01
Version: v1.0
OperatorRail:
- 5-Box planning
- paragraph execution checklist
- timed writing stability
- rewrite weakest paragraph
ArchitectRail (Sandbox only unless stable):
- bounded “unorthodox combination” drills
- fascination pull → richer detail
- fenced by OperatorSlip + LoadStability
Router:
If LoadStabilityIndex < T3:
  OperatorRail only
Else:
  10–20% ArchitectRail in Sandbox Window

ID: EDU_ENGINE_MATH_01
Version: v1.0
OperatorRail:
- method rails
- error taxonomy loops
- timed sets (speed rails)
ArchitectRail:
- variation ladder generator (change form, keep invariants)
Goal:
- TransferScore ↑ without TimePerMark collapse
Fence:
If TransferScore ↑ but TimePerMark worsens:
  truncate variation
  stitch speed rails
  retest

ID: EDU_ENGINE_SCI_01
Version: v1.0
OperatorRail:
- answer frames
- data interpretation drills
- timing stability
ArchitectRail:
- explanation corridors (why/how)
- hypothesis play (bounded)

ID: PARENT_RAILS_01
Version: v1.0
Contract:
Parents help without becoming tutors.
Tools:
- 10-min daily check-in (binary completion)
- “Show me your plan” prompt (PlanningCompleteness)
- weekly retest day (same sensors)
- protect sleep + routine rails (capacity S_cap)

ID: BTT_PROJECTION_LINE_01
Version: v1.0
Anchor:
PLACE:SGP:BTM
INST:SGP:BTM:BTT
ExportOutputs:
- Protocols (Diagnose/Repair/Retest)
- Sensor packs (EDU + AVOO)
- Pack registry schemas
- Z5 templates (countries/cities)
Principle:
Export schemas + adapters, not “claims”.

ID: EDU_Z5_COUNTRY_TEMPLATE_01
Version: v1.0
Fields:
- PLACE:ISO3
- ExamPhases (age bands)
- SkillFamilies (ENG/MATH/SCI)
- AdapterSlots (local topic names → universal skills)
- Default 12-week plan import
- SensorDict unchanged (only thresholds tuned)

ID: NEG_VOID_LIBRARY_01
Version: v1.0
NV_ENG_01: Why writing collapses under timing
NV_MATH_01: Why accuracy drops as topics increase
NV_MATH_02: Why memorised methods fail on variants
NV_SCI_01: Why answers are correct-but-low-marks
NV_META_01: Why motivation disappears after setbacks (capacity collapse)
Each page ends with:
Sensors triggered → Truncate → Stitch → Retest → Re-enter stable band

ID: RUNTIME_FOOTER_BTT_01
Version: v1.2
Phase: P0–P3
Zoom: Z0–Z6
Sensors: SENSOR_DICT_BTT_01 + SENSOR_DICT_AVOO_01
Repair: BTT_MODULE_ERCO_01
Schedule: BTT_MODULE_CHRONOHELM_01
AVOO: BTT_MODULE_AVOO_ACADEMY_01
Failure Trace:
  Z0 leak → Z1 collapse → Z3 instability → P0 fracture → repair → retest → P2 stability
Next Action:
  Diagnose → Repair → Retest

ID: RUNNER_BTT_01
Version: v1.0
Step 1: Run Backtest (triage or deep)
Step 2: Record sensors S1–S10 + A1–A7
Step 3: ERCO outputs RepairPlan + WindowPlan
Step 4: ChronoHelm schedules 12-week plan
Step 5: Execute:
  - Freeze (Operator) until stable
  - Sandbox (Architect) only if fenced
  - Deploy (exam-safe)
Step 6: Retest weekly with same sensors
Step 7: Promote only when stability holds under timing

ID: DEFLOCK_BTT_RUNTIME_01
Version: v1.2
BTT Runtime:
BukitTimahTutor.com is a running Education hub that converts:
Diagnose → Repair → Retest → Stabilise (P0→P3),
and exports its protocols as templates (Z5).
EducationOS Core Law (Rate Dominance):
Learning improves when Ġ_repair > Ḋ_decay under real load.
AVOO:
Role lattice for capability building:
Architect (generate) → Visionary (aim) → Oracle (predict) → Operator (execute)
Ground Zero:
INST:SGP:BTM:BTT is the canonical testbed where AVOO packs are installed,
measured under load, fenced for safety, then exported as templates.

ID: DIAG_BTT_ONE_PANEL_01
INPUTS
  StudentState + Goal + TimeBudget
  BacktestResults (accuracy/speed/transfer/load)
  SensorDict (EDU + AVOO)
PROCESS (Closed Loop)
  1) Diagnose (Backtest → leak map)
  2) ERCO decides (Truncate/Stitch + Windows)
  3) ChronoHelm schedules (weekly plan + ramps)
  4) Execute (Operator rails + bounded Architect sandbox)
  5) Retest (same sensors, same thresholds)
  6) Promote (difficulty + transfer) OR revert (truncate/stitch)
OUTPUTS
  RepairPlan + RetestSchedule + WindowPlan
  Subject Engines (ENG/MATH/SCI)
  AVOO Academy Packs (A/V/O/O)
  Projection Templates (BTM → World)

ID: INSTALL_BTT_01
Version: v1.2
InstallOrder:
1) BTT Control Tower (this page)
2) Diagnosis / Backtest Engine
3) ERCO Repair Overlay (decide + enforce windows)
4) ChronoHelmAI Scheduler (sequence + guardrails)
5) AVOO Academy (Ground Zero packs + choice budget)
6) Subject Engines:
   - English Engine (Composition + Language)
   - Math Engine (Primary/Secondary/A-Math)
   - Science Engine
7) Parent Toolkit (minimum-effort support rails)
8) Global Projection Templates (Z5 shells)

ID: BTT_ID_PACK_01
PLACE:SGP:BTM
INST:SGP:BTM:BTT
LANE:EDU
LANE:VOCAB
LANE:MATH
LANE:SCI
LANE:MIND
LANE:META

ID: SENSOR_DICT_BTT_01
Version: v1.0
S1  AccuracyRate (by topic + question-type)
S2  TimePerMark (speed stability)
S3  ErrorTaxonomy (careless / concept / misread / method / language)
S4  RetentionHalfLife (days-to-decay)
S5  TransferScore (unseen variants)
S6  LoadStabilityIndex (timed variance)
S7  PanicSpikeIndex (freeze/avoidance)
S8  PlanningCompleteness (before writing/solving)
S9  RevisionDepth (quality delta after rewrite)
S10 HelpSeekingLatency (delay-to-fix)
Thresholds (default bands; tune by level):
T1 AccuracyFloor = 0.70
T2 TransferFloor = 0.60
T3 LoadFloor = stable timed performance (variance capped)
T4 TimePerMark = within target band

ID: SENSOR_DICT_AVOO_01
Version: v1.0
A1 RoleMix (A/V/O/O proportions)
A2 ChoicePressureRatio ρ = S_inj / S_cap
A3 NoveltyBandwidth (distance from standard forms)
A4 FascinationPullIndex (curiosity pull)
A5 CorridorRetention (48h retention of new corridor)
A6 OperatorSlipRate (error inflation after novelty)
A7 PhaseShearProxy (variance under timing)
Thresholds:
T_A2  ρ_max_deploy = 1.0
T_A6  Slip_max = +15% vs baseline
T_A7  Shear_max = level-tuned cap

ID: FENCE_BTT_01
Version: v1.0
Fence Rule:
If (LoadStabilityIndex < T3) OR (ρ > T_A2) OR (Shear > T_A7):
  → TRUNCATE novelty/difficulty
  → STITCH fundamentals + rails
  → RETEST
  → Resume ramp only after stability returns
Windows:
Freeze Window (Operator Stability):
  - zero novelty
  - accuracy + speed rails
  - timed stability
Sandbox Window (Architect Exploration):
  - bounded novelty tasks only
  - short cycles, reversible
  - allowed only if OperatorSlip <= T_A6
Deploy Window (Exam Execution):
  - proven patterns only
  - timed sets + checklists
  - disallow untested novelty

ID: BTT_MODULE_CONTROL_TOWER_01
Version: v1.2
Contract:
- Provide install order + routing to all engines.
- Keep one canonical sensor dictionary.
- Provide “run loop” instructions for humans + LLMs.
Inputs:
- StudentProfile
- Goal + Timeline
- BacktestResults
- SENSOR_DICT_BTT_01 + SENSOR_DICT_AVOO_01
Outputs:
- RepairPlanRouter
- WindowPlanRouter
- EngineRouter (ENG/MATH/SCI + AVOO packs)
- ProjectionIndex (BTM → World templates)

ID: BTT_MODULE_BACKTEST_01
Version: v1.0
Contract:
Convert confusion into a measurable leak map + next actions.
Modes:
- TriageScan (15 min)
- DeepBacktest (45–60 min)
Outputs:
- LeakMap:
  Z0 Mistake patterns
  Z1 Topic weaknesses
  Z2 Habit/routine gaps
  Z3 Load failure triggers
- Baselines for sensors S1–S10
- Recommended WindowPlan entry state (Freeze/Sandbox/Deploy)

ID: BTT_MODULE_ERCO_01
Version: v1.0
Contract:
Choose actions using sensors + thresholds and enforce windows.
Inputs:
- LeakMap
- SensorReadings
- TimeBudget
Outputs:
- RepairPlan:
  - TruncationActions
  - StitchingActions
  - RetestSchedule
- WindowPlan:
  - Freeze / Sandbox / Deploy durations
- EscalationRules (when to promote difficulty)

ID: BTT_MODULE_CHRONOHELM_01
Version: v1.0
Contract:
Sequence RepairPlan into weeks/days with guardrails.
Outputs:
- WeeklyPlan (12-week default)
- DailyMinimum (non-negotiable)
- LoadRamp (week 1 → week 12)
- StopLoss rules:
  if 2 consecutive bad cycles → auto revert to Freeze + Stitch

ID: BTT_MODULE_AVOO_ACADEMY_01
Version: v1.0
Anchor:
  PLACE:SGP:BTM
  INST:SGP:BTM:BTT
Contract:
Train AVOO roles as executable skills under load,
fenced by Choice Budget so Operator stability is never sacrificed.
Inputs:
- StudentProfile (Level, Phase, Timeline)
- Domain (ENG/MATH/SCI/VOCAB/MIND/META)
- SensorDict EDU + AVOO
- Backtest baselines
Outputs:
- RoleMixPlan (A/V/O/O this week)
- ChoiceBudget (ρ constraints)
- TaskDeck (drills + prompts + timed sets)
- WindowPlan (Freeze/Sandbox/Deploy)
- ExportPack template (Z5)

ID: AVOO_RULE_CHOICE_BUDGET_01
Version: v1.0
Definitions:
S_inj = novelty/choices injected per unit time
S_cap = capacity (attention + working memory + routines)
ρ = S_inj / S_cap
Rule:
If (P ≤ P1) OR (LoadStabilityIndex < T3):
  - Architect tasks only in Sandbox Window
  - Operator dominates Deploy Window
Else (P2–P3 stable):
  - Expand Architect tasks gradually
  - Promote only if OperatorSlip and Shear remain below thresholds

ID: AVOO_PACK_REGISTRY_01
Version: v1.0
PACKS (Ground Zero):
PACK_A_ENG_01  Architect of Language (bounded novelty corridors)
PACK_O_ENG_01  Operator English Execution (timed writing + checklists)
PACK_A_MATH_01 Architect of Math (variation ladder, invariant-preserving)
PACK_O_MATH_01 Operator Math Execution (method rails + speed rails)
PACK_A_SCI_01  Architect of Science (explanation corridors, hypothesis play)
PACK_O_SCI_01  Operator Science Execution (answer frames under time)
PACK_META_01   Universal Architect Pack (transfer across domains)
PACK_META_02   Oracle Pack (prediction + error forecasting)
PACK_META_03   Visionary Pack (goal + constraint targeting)

PACK_SCHEMA_v1:
- Definition
- Mechanism
- TaskDeck (Z0 drills / Z1 routines / Z2 cadence)
- Sensors + thresholds
- Failure trace
- Repair loop (truncate/stitch/retest)
- Export template (Z5)

ID: EDU_ENGINE_ENG_01
Version: v1.0
OperatorRail:
- 5-Box planning
- paragraph execution checklist
- timed writing stability
- rewrite weakest paragraph
ArchitectRail (Sandbox only unless stable):
- bounded “unorthodox combination” drills
- fascination pull → richer detail
- fenced by OperatorSlip + LoadStability
Router:
If LoadStabilityIndex < T3:
  OperatorRail only
Else:
  10–20% ArchitectRail in Sandbox Window

ID: EDU_ENGINE_MATH_01
Version: v1.0
OperatorRail:
- method rails
- error taxonomy loops
- timed sets (speed rails)
ArchitectRail:
- variation ladder generator (change form, keep invariants)
Goal:
- TransferScore ↑ without TimePerMark collapse
Fence:
If TransferScore ↑ but TimePerMark worsens:
  truncate variation
  stitch speed rails
  retest

ID: EDU_ENGINE_SCI_01
Version: v1.0
OperatorRail:
- answer frames
- data interpretation drills
- timing stability
ArchitectRail:
- explanation corridors (why/how)
- hypothesis play (bounded)

ID: PARENT_RAILS_01
Version: v1.0
Contract:
Parents help without becoming tutors.
Tools:
- 10-min daily check-in (binary completion)
- “Show me your plan” prompt (PlanningCompleteness)
- weekly retest day (same sensors)
- protect sleep + routine rails (capacity S_cap)

ID: BTT_PROJECTION_LINE_01
Version: v1.0
Anchor:
PLACE:SGP:BTM
INST:SGP:BTM:BTT
ExportOutputs:
- Protocols (Diagnose/Repair/Retest)
- Sensor packs (EDU + AVOO)
- Pack registry schemas
- Z5 templates (countries/cities)
Principle:
Export schemas + adapters, not claims.

ID: EDU_Z5_COUNTRY_TEMPLATE_01
Version: v1.0
Fields:
- PLACE:ISO3
- ExamPhases (age bands)
- SkillFamilies (ENG/MATH/SCI)
- AdapterSlots (local topic names → universal skills)
- Default 12-week plan import
- SensorDict unchanged (only thresholds tuned)

ID: NEG_VOID_LIBRARY_01
Version: v1.0
NV_ENG_01: Why writing collapses under timing
NV_MATH_01: Why accuracy drops as topics increase
NV_MATH_02: Why memorised methods fail on variants
NV_SCI_01: Why answers are correct-but-low-marks
NV_META_01: Why motivation disappears after setbacks (capacity collapse)
Each page ends with:
Sensors triggered → Truncate → Stitch → Retest → Re-enter stable band

ID: RUNTIME_FOOTER_BTT_01
Version: v1.2
Phase: P0–P3
Zoom: Z0–Z6
Sensors: SENSOR_DICT_BTT_01 + SENSOR_DICT_AVOO_01
Repair: BTT_MODULE_ERCO_01
Schedule: BTT_MODULE_CHRONOHELM_01
AVOO: BTT_MODULE_AVOO_ACADEMY_01
Failure Trace:
  Z0 leak → Z1 collapse → Z3 instability → P0 fracture → repair → retest → P2 stability
Next Action:
  Diagnose → Repair → Retest

ID: RUNNER_BTT_01
Version: v1.0
Step 1: Run Backtest (triage or deep)
Step 2: Record sensors S1–S10 + A1–A7
Step 3: ERCO outputs RepairPlan + WindowPlan
Step 4: ChronoHelm schedules 12-week plan
Step 5: Execute:
  - Freeze (Operator) until stable
  - Sandbox (Architect) only if fenced
  - Deploy (exam-safe)
Step 6: Retest weekly with same sensors
Step 7: Promote only when stability holds under timing

ID: BTT_MODULE_BACKTEST_01
Version: v1.0
Contract:
Convert confusion into a measurable leak map + next actions.
Outputs:
- LeakMap (Z0–Z3)
- Sensor baselines (S1–S10 + A1–A7)
- Recommended WindowPlan entry state

ID: BTT_BACKTEST_OUTPUT_01
LeakMap:
Z0: Mistake patterns (misread, careless, method slip, language)
Z1: Topic weaknesses (which topics fail)
Z2: Routine gaps (planning, checking, revision depth)
Z3: Load triggers (timing, anxiety, multi-step fatigue)
Baselines:
S1–S10 + A1–A7

Rule:
If LoadStabilityIndex < threshold OR ChoicePressureRatio ρ > 1.0:
  Start in Freeze Window
Else:
  Allow Sandbox Window (bounded) + Deploy Window

ID: BTT_BACKTEST_RUNNER_01
Steps:
1) Run triage or deep backtest
2) Record S1–S10 + A1–A7
3) Produce LeakMap Z0–Z3
4) Hand off to ERCO for RepairPlan + Windows

ID: BTT_MODULE_ERCO_01
Version: v1.0
Contract:
Choose actions using sensors + thresholds and enforce windows.
Inputs:
LeakMap + SensorReadings + TimeBudget
Outputs:
RepairPlan + RetestSchedule + WindowPlan + EscalationRules

Freeze Window:
- Operator stability only
- no novelty injection
Sandbox Window:
- bounded Architect tasks
- only if OperatorSlip <= threshold
Deploy Window:
- exam-safe patterns only
- timed execution

ID: BTT_REPAIRPLAN_TEMPLATE_01
Truncate:
- (what is removed/reduced)
Stitch:
- (what is rebuilt)
- drills (Z0)
- routines (Z1)
- weekly cadence (Z2)
Retest:
- date + test type + thresholds
Windows:
- Freeze: __ days
- Sandbox: __ mins/day
- Deploy: __ sets/week
StopLoss:
- If 2 bad cycles → revert to Freeze + Stitch

ID: BTT_MODULE_CHRONOHELM_01
Version: v1.0
Contract:
Sequence RepairPlan into weeks/days with guardrails.
Outputs:
WeeklyPlan + DailyMinimum + LoadRamp + StopLoss

If 2 consecutive retests fail thresholds:
  Truncate difficulty/novelty
  Return to Freeze + Stitch
  Retest again

ID: BTT_CHRONOHELM_OUTPUT_01
WeeklyPlan:
- days, tasks, windows
DailyMinimum:
- 3 components
LoadRamp:
- week-by-week difficulty + timing increments

ID: BTT_MODULE_AVOO_ACADEMY_01
Version: v1.0
Anchor:
PLACE:SGP:BTM
INST:SGP:BTM:BTT
Contract:
Train AVOO roles as executable skills under load,
fenced by Choice Budget so Operator stability is never sacrificed.

ID: AVOO_RULE_CHOICE_BUDGET_01
ρ = S_inj / S_cap
If (P ≤ P1) OR (LoadStabilityIndex < threshold):
  Architect tasks only in Sandbox Window
  Operator dominates Deploy Window

ID: AVOO_PACK_REGISTRY_01
PACK_A_ENG_01, PACK_O_ENG_01
PACK_A_MATH_01, PACK_O_MATH_01
PACK_A_SCI_01, PACK_O_SCI_01
PACK_META_01, PACK_META_02, PACK_META_03

ID: AVOO_EXPORT_Z5_TEMPLATE_01
ExportUnit:
- PackSchema
- SensorDict
- Threshold tuning only
- AdapterSlots for local topic names

ID: EDU_ENGINE_ENG_01
Version: v1.0
OperatorRail:
- 5-Box planning
- execution checklist
- timed stability
- rewrite weakest paragraph
ArchitectRail:
- bounded novelty corridors
- fascination pull → richer detail
- sandbox only unless stable

ID: ENG_CHECKLIST_LOAD_01
1) Plan (5-box)
2) Topic sentences clear
3) Evidence/detail added
4) Linking words correct
5) Final check: tense + punctuation

ID: EDU_ENGINE_MATH_01
Version: v1.0
OperatorRail:
- method rails
- error taxonomy loops
- timed sets (speed rails)
ArchitectRail:
- variation ladder (change form, keep invariants)
Goal:
- TransferScore ↑ without speed collapse

ID: MATH_ERROR_TAX_01
- Misread
- Careless
- Wrong method
- Incomplete steps
- Weak fundamentals

ID: BTT_PROJECTION_LINE_01
Anchor:
PLACE:SGP:BTM
INST:SGP:BTM:BTT
ExportOutputs:
- Diagnose/Repair/Retest protocol
- Sensor packs (EDU + AVOO)
- Pack registry schemas
- Z5 templates (countries/cities)

ID: EDU_Z5_COUNTRY_TEMPLATE_01
Fields:
- PLACE:ISO3
- ExamPhases (age bands)
- SkillFamilies (ENG/MATH/SCI)
- AdapterSlots (local topic names → universal skills)
- Default 12-week plan import
- SensorDict unchanged (only thresholds tuned)

### Mechanism
* Architect energy = **controlled novelty** that makes readers *want to know more*.
* The constraint is **clarity + stability**.
* This pack is not “creative writing”. It is **corridor engineering**.
### TaskDeck

### Sensors + Thresholds

### Failure Trace

### Repair Loop
* **Truncate** novelty bandwidth (A3)
* **Stitch**: clarity rules (shorter sentences, fewer modifiers)
* **Retest** with timed paragraph execution
### Export Template (Z5)

---
# PACK_O_ENG_01 — Operator English Execution (Timed Writing + Checklists)
**Slug:** `pack-o-eng-01-operator-english-execution-v1-0`
**Category:** AVOO Academy → Operator Packs
## PACK_O_ENG_01 — Operator English Execution v1.0
### Win under timing (Deploy-safe)
### Definition

### Mechanism
* Operator = repeatable execution.
* This pack builds **timed stability** and reduces variance.
### TaskDeck

### Sensors + Thresholds

### Failure Trace

### Repair Loop
* **Truncate** length/time temporarily
* **Stitch** planning + paragraph frame
* **Retest** weekly timed composition
### Export Template
* Works everywhere: only tune timing and rubrics.
---
# PACK_A_MATH_01 — Architect of Math (Variation Ladder, Invariant-Preserving)
**Slug:** `pack-a-math-01-architect-of-math-v1-0`
**Category:** AVOO Academy → Architect Packs
## PACK_A_MATH_01 — Architect of Math v1.0
### Build transfer without destabilising speed
### Definition

### Mechanism
* Students fail not from ignorance, but from **form-shift**.
* Architect drills generate variants that keep the same core structure.
### TaskDeck

### Sensors + Thresholds

### Failure Trace

### Repair Loop
* **Truncate** Ladder to L1–L2
* **Stitch** method rails + timed drills
* **Retest** with unseen variant set
### Export Template
* Tune ladder difficulty per curriculum; keep invariants rule fixed.
---
# PACK_O_MATH_01 — Operator Math Execution (Method Rails + Speed Rails)
**Slug:** `pack-o-math-01-operator-math-execution-v1-0`
**Category:** AVOO Academy → Operator Packs
## PACK_O_MATH_01 — Operator Math Execution v1.0
### Accuracy + speed under exam constraints (Deploy-safe)
### Definition

### Mechanism
* Operator success = **rails**:
  * method rail (steps)
  * checking rail (common traps)
  * speed rail (time band)
### TaskDeck

### Sensors + Thresholds

### Failure Trace

### Repair Loop
* **Truncate** set length/time
* **Stitch** rails + micro-drills
* **Retest** weekly
### Export Template
* Universal; only adjust time bands and topic mixes.
---
# PACK_A_SCI_01 — Architect of Science (Explanation Corridors + Hypothesis Play)
**Slug:** `pack-a-sci-01-architect-of-science-v1-0`
**Category:** AVOO Academy → Architect Packs
## PACK_A_SCI_01 — Architect of Science v1.0
### Build deep explanations without breaking exam stability
### Definition

### Mechanism
* Many students memorise but can’t explain.
* Architect corridor = “why/how” reasoning with strict bounds.
### TaskDeck

### Sensors + Thresholds

### Failure Trace

### Repair Loop
* **Truncate** explanation length
* **Stitch** answer frames
* **Retest** with timed short answers
### Export Template
* Tune answer frames to local marking style; keep “bounded why/how” rule.
---
# PACK_O_SCI_01 — Operator Science Execution (Answer Frames Under Time)
**Slug:** `pack-o-sci-01-operator-science-execution-v1-0`
**Category:** AVOO Academy → Operator Packs
## PACK_O_SCI_01 — Operator Science Execution v1.0
### Score with clear frames, not long stories (Deploy-safe)
### Definition

### Mechanism
* Marks come from **points** and **keywords**, not length.
* Operator success = frame + clarity + speed.
### TaskDeck

### Sensors + Thresholds

### Failure Trace

### Repair Loop
* **Truncate** question count
* **Stitch** frames + keyword bank
* **Retest** timed short answers
### Export Template
* Universal frames; tune keyword lists per syllabus.
---
# PACK_META_01 — Universal Architect Pack (Transfer Across Domains)
**Slug:** `pack-meta-01-universal-architect-v1-0`
**Category:** AVOO Academy → META Packs
## PACK_META_01 — Universal Architect v1.0
### Generate corridors safely in any subject
### Definition

### Mechanism
* Transfer skill = being able to **move** while preserving structure.
* This pack trains that skill directly.
### TaskDeck

### Sensors + Thresholds

### Failure Trace

### Repair Loop
* **Truncate** change count
* **Stitch** rails
* **Retest** timed execution
---
# PACK_META_02 — Oracle Pack (Prediction + Error Forecasting)
**Slug:** `pack-meta-02-oracle-prediction-forecasting-v1-0`
**Category:** AVOO Academy → META Packs
## PACK_META_02 — Oracle Pack v1.0
### Predict failures before they happen (stability upgrade)
### Definition

### Mechanism
* Oracle skill = “I can see the failure coming.”
* It turns mistakes into *preventable events*.
### TaskDeck

### Sensors + Thresholds

### Failure Trace

### Repair Loop
* **Stitch** error atlas + pre-flight checklist
* **Retest** with same set types
---
# PACK_META_03 — Visionary Pack (Goal + Constraint Targeting)
**Slug:** `pack-meta-03-visionary-goal-constraint-targeting-v1-0`
**Category:** AVOO Academy → META Packs
## PACK_META_03 — Visionary Pack v1.0
### Set the aim correctly so effort becomes efficient
### Definition

### Mechanism
* Visionary = correct targeting under limited time.
* Prevents waste and “random grinding”.
### TaskDeck

### Sensors + Thresholds

### Failure Trace

### Repair Loop
* **Stitch** weekly target sheet
* **Retest** to validate target choice
* **Update** ChronoHelm schedule
---
## Universal Runtime Footer (paste into each pack page)

### What you see
* “I know what to write, but I can’t finish.”
* Good ideas, weak structure.
* Rushed ending.
* Grammar errors appear only when timed.
### Sensor Triggers (must measure)

### Failure Trace (mechanical)

### TRUNCATE (stop the spiral)

### STITCH (repair pipeline)

### RETEST (prove stability)

### Routing (what to run next)
* Run: `/btt-diagnosis-backtest-engine-v1-0/`
* Then: `/btt-erco-repair-overlay-v1-0/`
* Then: `/english-control-tower-v1-0/`
* Optional Sandbox: `PACK_A_ENG_01` only after stability
---
# NV_MATH_01 — Why Accuracy Drops As Topics Increase (Negative-Void)
**Slug:** `nv-math-01-why-accuracy-drops-when-topics-increase-v1-0`
**Category:** Negative Void Library → Math
## NV_MATH_01 — Why Accuracy Drops When Topics Increase v1.0
### The hidden failure is routing: wrong method selection under load
### Definition (Negative-Void)

### What you see
* “I understand, but still wrong.”
* Random switching of methods.
* Same student: good on homework, worse in mixed papers.
### Sensor Triggers

### Failure Trace

### TRUNCATE

### STITCH

### RETEST

### Routing
* Diagnose: `/btt-diagnosis-backtest-engine-v1-0/`
* Repair: `/btt-erco-repair-overlay-v1-0/`
* Execute: `/math-control-tower-v1-0/` + `PACK_O_MATH_01`
* Reopen Architect: `PACK_A_MATH_01` only after stability
---
# NV_MATH_02 — Why Memorised Methods Fail On Variants (Negative-Void)
**Slug:** `nv-math-02-why-memorised-methods-fail-on-variants-v1-0`
**Category:** Negative Void Library → Math
## NV_MATH_02 — Memorised Methods Fail On Variants v1.0
### The missing component is invariants — the “same concept” is not recognised
### Definition (Negative-Void)

### What you see
* Perfect on standard worksheets, fails unseen questions.
* “This one looks different” freeze.
### Sensor Triggers

### Failure Trace

### TRUNCATE

### STITCH

### RETEST

### Routing
* Execute: `PACK_A_MATH_01` (Sandbox) + immediate Deploy check using `PACK_O_MATH_01`
* Fence rule remains on (ρ ≤ 1.0 in Deploy)
---
# NV_SCI_01 — Why Answers Are Correct-But-Low-Marks (Negative-Void)
**Slug:** `nv-sci-01-why-correct-answers-get-low-marks-v1-0`
**Category:** Negative Void Library → Science
## NV_SCI_01 — Correct Ideas, Low Marks v1.0
### Missing link: answer frames + marking-point alignment
### Definition (Negative-Void)

### What you see
* Student explains verbally well, but written marks are low.
* Answers are vague or missing key terms.
* Too long, too slow.
### Sensor Triggers

### Failure Trace

### TRUNCATE

### STITCH

### RETEST

### Routing
* Execute: `/science-control-tower-v1-0/` + `PACK_O_SCI_01`
* Sandbox later: `PACK_A_SCI_01` bounded and fenced
---
# NV_META_01 — Why Motivation Disappears After Setbacks (Negative-Void)
**Slug:** `nv-meta-01-why-motivation-disappears-after-setbacks-v1-0`
**Category:** Negative Void Library → META
## NV_META_01 — Motivation Collapse After Setbacks v1.0
### The real failure is capacity collapse + missing retest loop
### Definition (Negative-Void)

### What you see
* Avoidance, procrastination, “I don’t want to do it.”
* Sudden dips after a bad test.
* Random practice without progress.
### Sensor Triggers

### Failure Trace

### TRUNCATE

### STITCH

### RETEST

### Routing
* Start: `/btt-diagnosis-backtest-engine-v1-0/`
* Repair: `/btt-erco-repair-overlay-v1-0/`
* META packs:
  * `PACK_META_02` (Oracle)
  * `PACK_META_03` (Visionary)
  * then re-open Architect tasks only when stable
---
## Universal Runtime Footer (paste into each Negative-Void page)

---
## 2) Phase Levels (P0–P3)

**Operational rule:**
If the student is in **P0/P1**, the system prioritises **Operator stability** and uses **Sandbox-only** Architect tasks.
---
## 3) Zoom Levels (Z0–Z6)

---
## 4) Core Loop: Diagnose → Repair → Retest

---
## 5) Truncation & Stitching (APRC Logic)

---
## 6) Windows: Freeze / Sandbox / Deploy

---
## 7) Education Sensors (S1–S10)

**Default thresholds (tune by level):**

---
## 8) AVOO Sensors (A1–A7)

**Thresholds:**

---
## 9) Choice Budget Law (ρ)

---
## 10) Modules (Canonical Runtime Blocks)
### 10.1 Control Tower

### 10.2 Backtest Engine

### 10.3 ERCO Repair Overlay

### 10.4 ChronoHelm Scheduler

### 10.5 AVOO Academy (Ground Zero)

---
## 11) Pack Schema v1 (Use on Every Pack Page)

---
## 12) Universal Runtime Footer (Paste Block)

---
## Operator Rail (Science): win marks under time
### The Operator Frame (universal)
Use this for most short answers:
**Claim → Evidence → Reason**
* **Claim:** the answer
* **Evidence:** what in the question/diagram supports it
* **Reason:** why it causes the outcome
### Daily Minimum (10–15 minutes)
* 5 short-answer questions timed
* Answer using the frame
* 2-minute keyword check
* Rewrite 2 weakest answers to match marking points
### Weekly Routine (2× per week)
* timed data interpretation set
* correction loop: rewrite answers to match marking points
**Pack page:** `PACK_O_SCI_01`
Slug: `/pack-o-sci-01-operator-science-execution-v1-0/`
---
## Architect Rail (Science): transfer and explanation corridors
This rail is powerful, but must be fenced.
### Architect tasks (bounded)
* **Why Chain:** 3-step why explanation (max 3 sentences)
* **Variable Swap:** change 1 variable; predict outcome
* **Diagram-to-words:** explain a diagram in 2 sentences
* **Hypothesis Play:** hypothesis → evidence → conclusion (3 lines)
**Pack page:** `PACK_A_SCI_01`
Slug: `/pack-a-sci-01-architect-of-science-v1-0/`
### When Architect Rail is allowed
* LoadStability is stable
* TimePerMark doesn’t inflate
* OperatorSlip does not rise beyond threshold
---
## Sensors that matter in Science (track weekly)
Copy block:

---
## Failure Pattern: “Correct ideas but low marks” (and the fix)
If the student is right but scores low, the failure is almost always one of these:
1. Missing keywords
2. Not writing marking points explicitly
3. Too long, too slow
4. Weak link between evidence and claim
### Repair actions
* truncate answer length
* stitch frames + keyword bank
* retest timed short answers weekly
**Negative Void page:**
`NV_SCI_01` → `/nv-sci-01-why-correct-answers-get-low-marks-v1-0/`
---
## Retest Protocol (Science)
Retest is not optional. It’s how improvement becomes real.
Weekly retest:
* short answers timed
* same frame
* compare:
  * keyword misses
  * time band
  * points achieved
  * variance under timing
Promotion rule:
* only increase difficulty/variation when stability holds.
---
## Copy Block: Science Engine (Almost-Code)

---
## Universal Runtime Footer (paste)

---
## Operator Rail (Math): rails + correction loops + speed band
### The Operator Method Rail (what it means)
A method rail is a repeatable sequence:
1. recognise question type
2. choose method
3. execute steps
4. check common traps
5. stay within time band
### Daily Minimum (10–15 minutes)
* 5-question micro-set timed
* mark immediately
* label every error (taxonomy)
* rewrite the correct method for wrong questions
### Weekly Routine (2× per week)
* 20–30 question mixed set timed
* correction loop: rewrite the 5 weakest questions properly
* 1 micro-retest midweek
**Pack page:** `PACK_O_MATH_01`
Slug: `/pack-o-math-01-operator-math-execution-v1-0/`
---
## Architect Rail (Math): Variation Ladder (fenced)
This rail builds transfer, but only after Operator stability exists.
### The Variation Ladder (3 levels)
* **L1:** same numbers, new wording
* **L2:** new numbers, same structure
* **L3:** new structure, same invariant
### Architect tasks (bounded)
* Invariant Spotter (what must remain true?)
* Form Flip (same invariant, new surface)
* Step Reorder Check (rail unchanged)
**Pack page:** `PACK_A_MATH_01`
Slug: `/pack-a-math-01-architect-of-math-v1-0/`
### When Architect Rail is allowed
* Accuracy is stable on standard sets
* TimePerMark is within band
* LoadStability variance is controlled
* OperatorSlip does not spike after novelty
---
## Sensors that matter in Math (track weekly)
Copy block:

---
## Two common Math collapse patterns (and fixes)
### Pattern A: Accuracy drops as topics increase
Cause: routing instability (wrong method selection under load)
Fix:
* truncate breadth
* stitch method rails
* stabilise with mixed micro-sets
* retest weekly
**Negative Void page:**
`NV_MATH_01` → `/nv-math-01-why-accuracy-drops-when-topics-increase-v1-0/`
### Pattern B: Memorised methods fail on variants
Cause: surface learning (invariants not recognised)
Fix:
* truncate ladder difficulty (L3→L1)
* stitch invariant spotting
* retest unseen variants weekly
**Negative Void page:**
`NV_MATH_02` → `/nv-math-02-why-memorised-methods-fail-on-variants-v1-0/`
---
## Retest Protocol (Math)
Retest is how stability becomes real.
Weekly retest (minimum):
* mixed micro-set timed
* compare:
  * wrong-method rate
  * time band
  * variance under timing
  * transfer score (when variants are active)
Promotion rule:
* only add difficulty/variants if stability holds.
---
## Copy Block: Math Engine (Almost-Code)

---
## Universal Runtime Footer (paste)

---
## Operator Rail (English): plan → execute → check
### The 5-Box Plan (must exist before writing)
Use a fixed plan that is fast:
1. **Beginning** (setup)
2. **Build-up** (problem grows)
3. **Climax** (peak moment)
4. **Resolution** (problem solved)
5. **Ending** (reflection / lesson)
Your plan doesn’t need beauty. It needs **structure**.
### Daily Minimum (10–15 minutes)
* 5 minutes: 5-box micro plan
* 8–10 minutes: one timed paragraph execution
* 2 minutes: quick checklist (tense/punct/linking)
### Weekly Routine (2× per week)
* one full timed composition
* rewrite the weakest paragraph (revision depth)
**Pack page:** `PACK_O_ENG_01`
Slug: `/pack-o-eng-01-operator-english-execution-v1-0/`
---
## Architect Rail (English): fascination corridors (fenced)
This rail is powerful, but only after Operator stability exists.
### Architect tasks (bounded)
* **Swap-2:** upgrade 2 bland words (keep meaning)
* **Contrast Pair:** one sentence with controlled contrast (X, but Y)
* **Verb Upgrade:** replace weak verbs with precise verbs
* **Hook Ladder:** write 3 hooks (mild → strong → strongest)
* **Detail Injector:** add 2 sensory details + 1 emotion cue (max 1 extra sentence)
**Pack page:** `PACK_A_ENG_01`
Slug: `/pack-a-eng-01-architect-of-language-v1-0/`
### When Architect Rail is allowed
* planning is stable
* timed variance is controlled
* OperatorSlip does not rise beyond threshold
* writing still finishes well under time
---
## Sensors that matter in English (track weekly)
Copy block:

---
## The most common English collapse pattern (and the fix)
### Pattern: Writing collapses under timing
Cause: plan disappears → structure collapses → time is lost → rushed ending
Fix:
* truncate complexity
* stitch 5-box plan + paragraph frames
* retest timed writing weekly
* reopen Architect tasks only after stability
**Negative Void page:**
`NV_ENG_01` → `/nv-eng-01-why-writing-collapses-under-timing-v1-0/`
---
## Retest Protocol (English)
Retest is how improvement becomes real.
Weekly retest (minimum):
* timed paragraph + timed full composition
* compare:
  * plan presence (S8)
  * finish rate
  * variance under time (S6)
  * rewrite improvement (S9)
Promotion rule:
* only add complexity or fascination corridors if stability holds.
---
## Copy Block: English Engine (Almost-Code)

---
## Universal Runtime Footer (paste)

---
## How to choose the right pack (fast routing)
### 1) Pick your Phase (P0–P3)
* **P0 (fracture):** performance collapse, avoidance, panic
* **P1 (unstable):** inconsistent, errors spike under timing
* **P2 (stable):** consistent on standard forms, improving
* **P3 (robust):** stable under timing + variation
**Rule:** If P0/P1, start with **Operator** packs and **Oracle** supports.
Architect packs are Sandbox-only until stable.
### 2) Pick your Goal
* **Accuracy under time**
* **Speed stability**
* **Transfer on variants**
* **Fascination + writing richness**
* **Prevent repeated mistakes**
* **Stop wasting effort (better targeting)**
### 3) Run pack → retest weekly
Every pack requires a weekly retest with the same sensors.
---
## Pack Library (Ground Zero)
### Operator Packs (Deploy-safe)
**Use these first when unstable, timed, or inconsistent.**
* **PACK_O_ENG_01 — Operator English Execution**
  `/pack-o-eng-01-operator-english-execution-v1-0/`
* **PACK_O_MATH_01 — Operator Math Execution**
  `/pack-o-math-01-operator-math-execution-v1-0/`
* **PACK_O_SCI_01 — Operator Science Execution**
  `/pack-o-sci-01-operator-science-execution-v1-0/`
---
### Architect Packs (Sandbox-only unless stable)
**Use these to build transfer, depth, and fascination — fenced by thresholds.**
* **PACK_A_ENG_01 — Architect of Language**
  `/pack-a-eng-01-architect-of-language-v1-0/`
* **PACK_A_MATH_01 — Architect of Math**
  `/pack-a-math-01-architect-of-math-v1-0/`
* **PACK_A_SCI_01 — Architect of Science**
  `/pack-a-sci-01-architect-of-science-v1-0/`
---
### META Packs (cross-subject)
**These make progress more efficient by preventing repeat failures and mis-aim.**
* **PACK_META_01 — Universal Architect (transfer skill)**
  `/pack-meta-01-universal-architect-v1-0/`
* **PACK_META_02 — Oracle (prediction + error forecasting)**
  `/pack-meta-02-oracle-prediction-forecasting-v1-0/`
* **PACK_META_03 — Visionary (goal + constraint targeting)**
  `/pack-meta-03-visionary-goal-constraint-targeting-v1-0/`
---
## “If you are X, start with Y” (simple routing rules)
### If performance collapsed suddenly (panic, avoidance, freeze)
Start:
1. Diagnose
2. ERCO
3. ChronoHelm
4. Operator pack only (Freeze window)
Choose:
* Math: `PACK_O_MATH_01`
* English: `PACK_O_ENG_01`
* Science: `PACK_O_SCI_01`
  Add: `PACK_META_02` (Oracle) if mistakes repeat.
---
### If student is okay at homework but fails mixed papers
Cause is usually routing instability under load.
Start:
* `PACK_O_MATH_01` + mixed micro-sets + correction loops
  Then reopen:
* `PACK_A_MATH_01` only after stability
Also read:
* `NV_MATH_01` and `NV_MATH_02`
---
### If student has correct ideas but low marks (Science)
Start:
* `PACK_O_SCI_01` (frames + keywords + timed short answers)
  Then reopen:
* `PACK_A_SCI_01` bounded
Also read:
* `NV_SCI_01`
---
### If writing collapses under timing
Start:
* `PACK_O_ENG_01` (plan + execute + checklist + rewrite)
  Then reopen:
* `PACK_A_ENG_01` bounded
Also read:
* `NV_ENG_01`
---
### If motivation disappears after setbacks
Start:
* `PACK_META_02` (Oracle) + `PACK_META_03` (Visionary)
  And run:
* daily minimum via ChronoHelm
  Keep Architect tasks Sandbox-only until stable
Also read:
* `NV_META_01`
---
## Sensors used by the Academy (how packs are fenced)
Copy block:

---
## Export (BTM → World) — pack templates
Once a pack works locally, it becomes exportable as a template:
* same tasks
* same sensors
* only tune thresholds and topic adapters
Go here:
* `/btt-global-projection-btm-to-world-v1-0/`
---
## Universal Runtime Footer (paste)