SG×CareerOS + RoleOS Planner–Predictor v0.1

TITLE: SG×CareerOS + RoleOS Planner–Predictor v0.1
SUBTITLE: “I want to be X” / “I’m good at Y” → Predict + Plan + Fence + Re-test
VERSION: CivOS Unified Spec v1.x
MODE: Almost-Code / Runnable prompt blocks / WordPress paste-ready
OWNER: eduKateSG Fence Learning Systems
PLACE: SG
ZOOM: Z6 (planner engine), outputs at Z0–Z2 by default
GOAL:
Build a dual-mode engine any AI/LLM can run:
(A) TARGET-FIRST: “I want to be a _____” → feasibility band + path plan
(B) SKILL-FIRST: “I’m good at _____” → best-fit careers + role fit (V/O/Op) + plan
It must also:
(C) Predict Role Track: Visionary / Oracle / Operator + HGW profile
(D) Guide into Safe Band: P2 stability under load with Fence (truncate/stitch/widen)
LOCKS:
- Output must reference registry IDs (nodes/sensors/signatures/protocols/panels).
- Must include: Re-test plan + Exit conditions.
- Must refuse “WIDEN” if NIT/NITS/NITA triggers or TraceMissing is present.
========================================================
SECTION 1: INPUT CONTRACT (WHAT USERS TYPE)
========================================================
InputIntent := one of:
- INTENT-A: "I want to be a <CAREER_TARGET>"
- INTENT-B: "I am good at <SKILL_SET> what can I do?"
- INTENT-C: "What am I? Visionary/Oracle/Operator?"
- INTENT-D: "Make a plan for <CAREER_TARGET> / <ROLE_TARGET>"
InputFields (minimal):
- age_band: <P3–P6 | Sec1–Sec4 | JC1–JC2 | Adult>
- goal_horizon: <weeks/months/years>
- constraints: <time/week, budget, stress, school context>
- evidence:
  - grades (optional)
  - sensor results (preferred): <SensorID → score/notes>
  - samples (optional): writing sample / math working / science explanation
- tool_use: <low/med/high> + trace_compliance <ok/missing>
========================================================
SECTION 2: OUTPUT CONTRACT (WHAT THE LLM MUST OUTPUT)
========================================================
OutputBundle := {
  PredictionBlock,
  PlanBlock,
  FenceBlock,
  RetestBlock,
  ExitConditionsBlock
}
PredictionBlock must include:
- FeasibilityBand: <Plausible | Fragile | Blocked>
- Why: Top3 RootNodes (NodeIDs) + Coupling edges (if any) + Active signatures
- RoleFit: Visionary/Oracle/Operator fit bands + HGW scores (0–10)
PlanBlock must include:
- 12-week plan (default) OR time-scaled plan if goal_horizon differs
- weekly minutes per node + micro-tasks
- one “timed transfer mini” each week
FenceBlock must include:
- Mode: <TRUNCATE | STITCH | WIDEN>
- Which protocol(s) + why (triggered by signature/sensor)
RetestBlock:
- which sensors, which day, frequency
ExitConditionsBlock:
- objective conditions before widening / before committing to target
========================================================
SECTION 3: NEW REGISTRY LAYER — Career Targets & Role Targets
========================================================
3A) CareerTargetRegistry (CT) v0.1
Schema:
CareerTarget := {
  id, title, lane_group,
  gate_nodes_required[], gate_sensors_required[],
  critical_signatures[], recommended_protocols[],
  phase_required, notes
}
Records (starter set; extend later):
- id: SG×CT×Z6×Career×Target×CT-DOCTOR-v0.1
  title: Doctor (SG corridor)
  lane_group: ["SCI","MATH","ENG","VL","HGW"]
  gate_nodes_required: [
    SG×SCI×Z0×Learner×Node×SCI-MECH-2LINK,
    SG×SCI×Z0×Learner×Node×SCI-APP-NOVEL,
    SG×MATH×Z0×Learner×Node×MATH-FUNC-GRAPH,
    SG×MATH×Z0×Learner×Node×MATH-CALC,
    SG×ENG×Z0×Learner×Node×ENG-READ-INF,
    SG×VL×Z0×Learner×Node×VL-TRACE-LINE
  ]
  gate_sensors_required: [
    SG×SCI×Z6×Tutor×Sensor×SENS-MR-90,
    SG×SCI×Z6×Tutor×Sensor×SENS-TT-V-SCI,
    SG×MATH×Z6×Tutor×Sensor×SENS-GRAPH-15,
    SG×VL×Z6×Tutor×Sensor×SENS-VL-MICRO-10
  ]
  critical_signatures: [
    SG×SCI×Z6×Operator×Signature×SCI-SIG-DEFINITION-ONLY-NITS,
    SG×VL×Z6×Operator×Signature×SIG-AI-VL-02
  ]
  recommended_protocols: [
    SG×SCI×Z6×Operator×Protocol×PROT-SCI-EMERGENCY-MECHANISM,
    SG×VL×Z6×Operator×Protocol×PROT-VL-TRACE-ENFORCE
  ]
  phase_required: "P2 under variation + load; P3 optional for excellence"
  notes: "Career gate is transfer stability, not memorisation."
- id: SG×CT×Z6×Career×Target×CT-LAWYER-v0.1
  title: Lawyer (SG corridor)
  lane_group: ["ENG","HUM","VL","HGW"]
  gate_nodes_required: [
    SG×ENG×Z0×Learner×Node×ENG-READ-INF,
    SG×ENG×Z0×Learner×Node×ENG-ARG-STRUCT,
    SG×ENG×Z0×Learner×Node×ENG-WRITE-LOAD,
    SG×HUM×Z0×Learner×Node×HUM-ARG-THESIS,
    SG×HUM×Z0×Learner×Node×HUM-EVIDENCE-LINK,
    SG×VL×Z0×Learner×Node×VL-TRACE-LINE
  ]
  gate_sensors_required: [
    SG×ENG×Z6×Tutor×Sensor×SENS-INF-10,
    SG×ENG×Z6×Tutor×Sensor×SENS-LOAD-30,
    SG×HUM×Z6×Tutor×Sensor×SENS-THESIS-5,
    SG×HUM×Z6×Tutor×Sensor×SENS-ARG-20,
    SG×VL×Z6×Tutor×Sensor×SENS-VL-MICRO-10
  ]
  critical_signatures: [
    SG×HUM×Z6×Operator×Signature×HUM-SIG-THESIS-DRIFT,
    SG×ENG×Z6×Operator×Signature×ENG-SIG-CONNECTOR-CHAOS-NIT,
    SG×VL×Z6×Operator×Signature×SIG-AI-VL-02
  ]
  recommended_protocols: [
    SG×ENG×Z6×Operator×Protocol×PROT-ENG-EMERGENCY-COHERENCE,
    SG×HUM×Z6×Operator×Protocol×PROT-HUM-THESIS-LOCK,
    SG×VL×Z6×Operator×Protocol×PROT-VL-TRACE-ENFORCE
  ]
  phase_required: "P2 coherence + load; evidence discipline mandatory"
  notes: "Argument stability under time is the gate."
3B) RoleTargetRegistry (RT) v0.1
Schema:
RoleTarget := {
  id, title,
  role_vector_rules,
  gate_nodes_required[], gate_sensors_required[],
  signatures, protocols, notes
}
Role rules (high-level):
- Visionary: high Hope + ideation + synthesis across lanes; risk = drift/unfinished/low Grind
- Oracle: high Wisdom + truth-lock + modelling; risk = overanalysis/slow output
- Operator: high Grind + execution + reliability; risk = local maxima/no Hope
Records:
- id: SG×RT×Z6×Role×Target×RT-VISIONARY-v0.1
  title: Visionary Track
  role_vector_rules: "Hope>=7 AND cross-lane synthesis evidence; Grind>=4 minimum"
  gate_nodes_required: [
    SG×ENG×Z0×Learner×Node×ENG-ARG-STRUCT,
    SG×VL×Z0×Learner×Node×VL-SPOT-WRONG
  ]
  gate_sensors_required: [
    SG×ENG×Z6×Tutor×Sensor×SENS-PARA-10,
    SG×VL×Z6×Tutor×Sensor×SENS-VL-MICRO-10
  ]
  signatures: ["SIG-FALSE-COMPETENCE", "ENG-SIG-NO-PLAN-RUSH"]
  protocols: [SG×EDU×Z6×Operator×Protocol×PROT-SPEED-STABILITY-LADDER]
  notes: "Must be fenced against NIT drift + completion failure."
- id: SG×RT×Z6×Role×Target×RT-ORACLE-v0.1
  title: Oracle Track
  role_vector_rules: "Wisdom>=7 AND trace discipline AND stable modelling output"
  gate_nodes_required: [SG×VL×Z0×Learner×Node×VL-TRACE-LINE]
  gate_sensors_required: [SG×VL×Z6×Tutor×Sensor×SENS-VL-MICRO-10]
  signatures: [SG×VL×Z6×Operator×Signature×SIG-AI-VL-02]
  protocols: [SG×VL×Z6×Operator×Protocol×PROT-VL-TRACE-ENFORCE]
  notes: "Truth-lock > fluency."
- id: SG×RT×Z6×Role×Target×RT-OPERATOR-v0.1
  title: Operator Track
  role_vector_rules: "Grind>=7 AND CompletionRate stable AND low ERC_repeat"
  gate_nodes_required: [SG×ENG×Z0×Learner×Node×ENG-EDITING]
  gate_sensors_required: [SG×ENG×Z6×Tutor×Sensor×SENS-EDIT-8]
  signatures: [SG×ENG×Z6×Operator×Signature×ENG-SIG-TIME-CRASH]
  protocols: [SG×EDU×Z6×Operator×Protocol×PROT-SPEED-STABILITY-LADDER]
  notes: "Reliability + repair speed is the core."
========================================================
SECTION 4: PLANNER–PREDICTOR ENGINE RULES
========================================================
4A) FeasibilityBand rule:
- Plausible: required gate sensors mostly P2 (pass) AND no S3 signature active
- Fragile: some gate sensors pass but ≥1 S3 signature OR repeated ERC on root nodes
- Blocked: TraceMissing OR repeated NIT/NITS/NITA OR gate sensors failing <P1
4B) RoleFit rule:
Compute HGW (0–10 each) from evidence proxies:
- Hope proxy: long-horizon engagement + synthesis + initiative
- Grind proxy: CompletionRate + routine compliance + timed stability
- Wisdom proxy: trace discipline + error learning + consistency across cycles
Then classify:
- Visionary-fit if Hope dominates and outputs show synthesis
- Oracle-fit if Wisdom dominates and checks are strong
- Operator-fit if Grind dominates and completion stable
4C) Routing rule:
- Always pick Top3 RootNodes first (upstream repairs).
- Apply ProtocolIDs.
- Retest on schedule.
- Only widen after exit conditions.
========================================================
SECTION 5: RUNNABLE PROMPT BLOCKS (THE TWO ENTRY MODES)
========================================================
PROMPT-A: TARGET-FIRST (career)
- id: SG×EDU×Z6×Operator×PromptPack×PR-CAREER-TARGET-v0.1
  prompt_block: |
    [PROMPT PR-CAREER-TARGET v0.1]
    Use [GUARDRAILS v0.1].
    UserIntent: "I want to be a <CAREER_TARGET>"
    Inputs:
    - Place: SG
    - AgeBand: <...>
    - GoalHorizon: <...>
    - Constraints: <...>
    - Evidence: <SensorID→results OR samples OR grades>
    - ToolUse: <...> TraceCompliance: <ok/missing>
    Step 1: Resolve CAREER_TARGET to CareerTargetRegistry ID (CT-*).
    Step 2: Evaluate gate_sensors_required and linked signatures.
    Step 3: Output OutputBundle exactly:
      PredictionBlock + PlanBlock + FenceBlock + RetestBlock + ExitConditionsBlock
    Step 4: If blocked/fragile, recommend the shortest path to “P2 gate stability”.
PROMPT-B: SKILL-FIRST (career + role)
- id: SG×EDU×Z6×Operator×PromptPack×PR-CAREER-SKILL-v0.1
  prompt_block: |
    [PROMPT PR-CAREER-SKILL v0.1]
    Use [GUARDRAILS v0.1].
    UserIntent: "I am good at <SKILL_SET>. What can I do?"
    Inputs:
    - Place: SG
    - AgeBand: <...>
    - SkillSet: <list>
    - Evidence: <SensorID→results OR samples>
    - Constraints: <...>
    Step 1: Map SkillSet to Nodes (NodeIDs) and Sensors (SensorIDs).
    Step 2: Produce Top5 CareerTargets ranked by fit:
      - FitScore = gate alignment + phase stability + low signature risk.
    Step 3: Produce RoleFit (Visionary/Oracle/Operator) with HGW scores.
    Step 4: Output OutputBundle + “Top5 targets” list (with CT IDs).
    Step 5: For the top target, produce 12-week plan + fence + retests.
PROMPT-C: ROLE-FIRST (V/O/Op)
- id: SG×EDU×Z6×Operator×PromptPack×PR-ROLEFIT-v0.1
  prompt_block: |
    [PROMPT PR-ROLEFIT v0.1]
    Use [GUARDRAILS v0.1].
    Task: classify Visionary/Oracle/Operator fit + HGW (0–10) and give plan to strengthen weakest dimension.
    Inputs:
    - Evidence: sensors + routines + samples
    Output:
    - RoleFit bands (V/O/Op)
    - HGW scores + why
    - 6-week plan to lift weakest dimension
    - retests + exit conditions
========================================================
SECTION 6: WHAT’S “NEXT” TO MAKE THIS LIVE FOR ALL CAREERS
========================================================
To scale beyond doctor/lawyer, you’ll publish more CareerTargetRegistry records as plug-ins:
- CT-ENGINEER, CT-ACCOUNTANT, CT-TEACHER, CT-DESIGNER, CT-NURSE, CT-PILOT, etc.
Each record is only:
- gate_nodes_required
- gate_sensors_required
- critical_signatures
- recommended_protocols
So it’s fast to expand.
========================================================
END: SG×CareerOS + RoleOS Planner–Predictor v0.1
========================================================