This is the control system: diagnose fast, tag failures (Z0–Z3), pick the smallest repair loop that moves the learner back toward P3 reliability.

RR.0 Repair Router Object (RRO)

ID: ENG.RR.RRO.v0_1

			
RRO = {
  Sample,                 # 150–200 words (or 2–3 min speaking)
  Context,                # PSLE | Secondary | Adult | Exam | Work
  PhaseEstimate,          # P0–P3 (by stability under variation)
  Tags: {Z0[], Z1[], Z2[], Z3[]},
  DominantLeakSet,        # top 1–3 tags by impact
  RepairPlan,             # 2-week micro plan (daily loops)
  LoadPlan,               # timed output schedule
  ReTestRule              # when & how to re-measure
}

		

RR.1 5-Minute Diagnostic Protocol (No guessing)

ID: ENG.RR.DIAG5.v0_1

Input

one writing sample (150–200 words) OR one paragraph + 6 sentences
optional: timed (10 minutes) if measuring load collapse

Step 1 — Phase Estimate (30s)

P0: cannot form stable sentences
P1: templates only; breaks when prompt changes
P2: meaning ok; frequent leaks; load drops under time
P3: stable under time + topic variation + self-corrects

Step 2 — Z0 Boundary First (60s)

Count:

run-ons/fragments
article/pronoun agreement leaks
preposition/connectors misuse

If FT.Z0.PUNCT_BOUNDARY is high → fix first (everything else depends on boundaries).

Step 3 — Z1 Coordinate Leaks (60s)

Check:

missing do-support
wrong verb form (did + Ved, has + Ved, modal + -s)
question inversion / negation placement

Step 4 — Z2 Spine (60s)

Does paragraph contain:

Claim? Reason? Evidence? Explanation? Link?
If missing R or X, tag it.

Step 5 — Z3 Fit (60s)

mode fit (narrative vs argument vs situational)
intro/conclusion present?
register mismatch?
pacing debt?

Output

Select top 1–3 Dominant tags only. (Don’t fix everything at once.)

RR.2 Unified Tag Set (Canonical)

ID: ENG.RR.TAGS.v0_1

RR.3 Dominant Leak Rules (what to fix first)

ID: ENG.RR.DOMINANCE_RULES.v0_1

Rule A — Boundary First
If FT.Z0.PUNCT_BOUNDARY high → repair before any Z1/Z2 training.

Rule B — Sentence Engine Next
If Z1 coordinate leaks are high, Z2/Z3 work will not hold under load.

Rule C — Explanation is the mark engine
If FT.Z2.NO_EXPLAIN or FT.Z2.EXAMPLE_DUMP → fix early (biggest score gain).

Rule D — Wrong Mode kills everything
If FT.Z3.WRONG_MODE (argument written as story, etc.) → fix before polishing language.

Rule E — Limit active repairs
Never run more than:

2 Z0 repairs + 1 Z1 drill + 1 Z2 template per week
(Too many = no consolidation → P2 stays leaky.)

RR.4 Repair Library Mapping (Tag → Loop)

ID: ENG.RR.MAP.v0_1

Z0 mappings

FT.Z0.PUNCT_BOUNDARY → RL.Z0.BOUNDARY
FT.Z0.ARTICLES + FT.Z0.COUNTABILITY → RL.Z0.ART3
FT.Z0.PRONOUN_CASE → RL.Z0.PROCASE
FT.Z0.AGREEMENT → RL.Z0.AGR10
FT.Z0.PREPOSITIONS → RL.Z0.PREP_MAP
FT.Z0.CONNECTORS → RL.Z0.CONN_LOGIC
FT.Z0.IRREGULAR → RL.Z0.IRR_TRI
FT.Z0.APOSTROPHE → RL.Z0.ITS
FT.Z0.HOMOPHONES → RL.Z0.HOMO
FT.Z0.COLLOCATIONS → RL.Z0.COLLOC

Z1 mappings

FT.Z1.DO_SUPPORT → ENG.Z1.DRILLGEN slice (NEG+QYN)
FT.Z1.VFORM → V0/Ved/Ven triangle + coordinate drills (PAST/COMPLETE)
FT.Z1.INVERSION → QYN/QWH drills
FT.Z1.NEG_ATTACH → “not attaches to first aux” drill
FT.Z1.AUX_CHAIN → have/be ordering drills (has been Ving, has been Ven)
FT.Z1.STATE_ILLEGAL → static-vs-dynamic constraints pack
FT.Z1.PASSIVE_BUILD → passive table drills
FT.Z1.MODAL_FORM → modal + V0 swaps

Z2 mappings

FT.Z2.NO_REASON → RL.Z2.BECAUSE3
FT.Z2.EXAMPLE_DUMP / FT.Z2.NO_EXPLAIN → RL.Z2.SOWHAT
FT.Z2.DRIFT → RL.Z2.DRIFT_LOCK
FT.Z2.CONNECTOR_MISMATCH → RL.Z2.CONN_SWAP
FT.Z2.REFERENCE_BREAK → RL.Z2.REF_CHAIN
FT.Z2.CHOPPY → RL.Z2.COMBINE
FT.Z2.WEAK_EVIDENCE → RL.Z2.SHOW3

Z3 mappings

FT.Z3.NO_THESIS / FT.Z3_NO_ROADMAP → RL.Z3.INTRO30
FT.Z3_WEAK_ENDING → RL.Z3.CONC30
FT.Z3_NO_ARC → RL.Z3.ARC6
FT.Z3_NO_COUNTER_REBUT → rebuttal ladder (RB1/RB2/RB3)
FT.Z3_REGISTER_MISMATCH → RL.Z3.REG3
FT.Z3_NO_TRANSITIONS → RL.Z3.BRIDGE
FT.Z3_PACING_DEBT → RL.Z3.STOPLOSS

RR.5 12–16 Week Upgrade Schedule (Education OS compatible)

ID: ENG.RR.SCHEDULE_16W.v0_1

Phase 1 (Weeks 1–4): Seal leaks (Z0 + Z1)

Daily: 1× Z0 loop (5–8 min)
3×/week: Z1 coordinate slice drills (10–12 min)
Weekly: 1 timed paragraph (10 min) + tag review

Goal: reduce illegal coordinates + boundary leaks.

Phase 2 (Weeks 5–8): Paragraph spine (Z2)

3×/week: CREXL template paragraph (one)
Daily: keep 1 Z0 loop (top leak)
Weekly: 2 paragraphs timed + “so-what” enforcement

Goal: explanation-after-evidence becomes automatic.

Phase 3 (Weeks 9–12): Discourse control (Z3)

2×/week: full piece outline (argument or story arc)
1×/week: full timed writing
Add: intro/conclusion speed drills

Goal: mode fit + coherence + pacing.

Phase 4 (Weeks 13–16): Transfer + load stability

full exam simulations (monthly / fortnightly)
targeted repairs only (don’t re-open many tags)
focus: stop-loss endings + drift control under speed

Goal: P3 reliability under exam load.

RR.6 Re-Test Rule (when to change plan)

ID: ENG.RR.RETEST.v0_1

Re-test every 7 days with a short sample.
Only change DominantLeakSet if:
top tag reduced by >60%, OR
a new tag becomes dominant under timing.

RR.7 “Failure Mode Trace” (CivOS-style schematic)

ID: ENG.RR.FAILTRACE.v0_1

Example trace:

			
Z0 boundary leak (run-ons) →
Z1 aux spine misbuilt (do-support missing) →
Z2 spine collapses (no explanation) →
Z3 pacing debt (weak ending) →
P2→P0 drop under time

		

Repair routing:

			
RL.Z0.BOUNDARY + Z1 NEG/QYN slice →
RL.Z2.SOWHAT + RL.Z3.STOPLOSS →
P2 stabilises → P3 transfer

Copy-Paste WordPress Spec Block (Repair Router)

ENG_OS.RR.v0_1

Diagnose in 5 minutes: Phase estimate → Z0 boundary → Z1 coordinate → Z2 spine → Z3 fit
Tag system: FT.Z0.* / FT.Z1.* / FT.Z2.* / FT.Z3.*
Dominance rules: boundary first; sentence engine before paragraph; explanation is mark engine; wrong mode kills everything
Plan: 12–16 weeks (seal leaks → spine → discourse → transfer/load)
Retest weekly; change plan only when dominance shifts

ENGLISH OS v0.1 — Unified Repair Router (Canonical)