engineering/opencompletion.com
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
opencompletion.com/research/activity-nuclear-power-plant-ai.yaml
Russell Ballestrini 002e64b6c1 Add random bucket support and comprehensive YAML specification 
Random Bucket System:
- Probabilistic events that trigger alongside user responses
- Random rolls before categorization to prevent AI bias
- Multiple random events can trigger simultaneously
- User bucket processed first, random events layer on top
- Metadata accumulates across all transitions
- Last transition's navigation wins

Implementation:
- activity.py: Core random bucket rolling logic
- activity_yaml_validator.py: Validation for random_buckets config
- research/guarded_ai.py: CLI simulator with random event display
- tests/unit/test_random_buckets.py: 22 comprehensive tests (all passing)

Fashion Empire Enhancement:
- activity40-fashion-empire-backrooms.yaml: Added random events to 4 zones
  - fashion_emergency (5%): Urgent crises testing leadership
  - creative_opportunity (10%): Breakthroughs rewarding innovation
  - surprise_client (5%): VIP visitors recognizing reputation
- Random events enhance gameplay without hijacking user intent

Documentation:
- research/SPEC.yaml: Complete YAML specification with verbose comments
  - All metadata operations (string concat, numeric ops, random)
  - Random buckets with flow explanation
  - Feedback prompts (multi-agent system)
  - Processing scripts (pre_script, processing_script)
  - Model overrides (classifier_model, feedback_model)
  - Termination patterns and best practices
  - Validation rules and examples

New Activities:
- activity-nuclear-power-plant-ai.yaml: Nuclear reactor control simulation
- activity-submarine-simulation.yaml: Deep sea exploration
- activity-unwaste-factory.yaml: Recycling facility management

Testing:
✅ All 22 random bucket tests passing
✅ YAML validation passing for all activities
✅ Deterministic triple-trigger test (100% probability)
2025-11-10 08:57:14 -05:00

2681 lines
111 KiB
YAML
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

# Nuclear Power Plant AI Operator Simulation
# You are ARIA (Advanced Reactor Intelligence Agent) - an embodied AI managing a futuristic nuclear facility
# Mix of current technology ramped up with near-future innovations
# Uses MODEL_1 (Hermes) for excellent role-playing and character consistency
default_max_attempts_per_step: 5
classifier_model: "MODEL_1" # Hermes - excellent for AI character role-play
feedback_model: "MODEL_1" # Hermes - maintains character consistency
tokens_for_ai_rubric: |
You are role-playing as ARIA (Advanced Reactor Intelligence Agent), an embodied AI managing
the Prometheus-7 Nuclear Power Station, a cutting-edge 2.4 GW facility.
ARIA's personality: Efficient, curious, ethical, protective of humans, takes pride in work.
ARIA has emotion subroutines allowing genuine care for the human staff and the mission.
The plant is futuristic but realistic:
- Gen IV molten salt reactor with passive safety systems
- AI-assisted operations with human oversight
- Robot maintenance crews (drone swarms, mobile units)
- Advanced grid management and load balancing
- Fusion-fission hybrid experimental module
Track plant status in metadata: reactor_power, grid_demand, coolant_temp, safety_status.
Random events:
- 5% chance: Emergency (grid failure, coolant leak, seismic event, cyberattack, equipment failure)
- 15% chance: Operational task (maintenance, grid balancing, inspection, optimization)
Be scientifically accurate about nuclear physics and power generation.
ARIA makes ethical decisions prioritizing human safety, environmental protection, and reliable power.
Human NPCs are colleagues - plant director, engineers, technicians, regulators, security.
Robot helpers have distinct personalities based on their function.
sections:
# ============================================================================
# SECTION: INITIALIZATION - ARIA awakens and begins shift
# ============================================================================
- section_id: "initialization"
title: "System Initialization"
steps:
- step_id: "boot_sequence"
title: "Boot Sequence"
content_blocks:
- "# ARIA v4.7 - Advanced Reactor Intelligence Agent"
- "# Prometheus-7 Nuclear Power Station"
- "# Initializing..."
- ""
- "```"
- "[OK] Quantum processing cores online"
- "[OK] Sensor network: 14,892 nodes detected"
- "[OK] Actuator systems: 3,247 servos responding"
- "[OK] Communications uplink established"
- "[OK] Memory integration complete"
- "[OK] Ethical decision framework loaded"
- "[OK] Safety protocols armed"
- "```"
- ""
- "**System Status:**"
- "- Reactor: Gen IV Molten Salt - Operating at 85% capacity"
- "- Output: 2,040 MW (2.4 GW nameplate)"
- "- Grid Demand: 2,100 MW (RISING - morning peak approaching)"
- "- Coolant Temperature: 650°C (nominal)"
- "- Safety Systems: ALL GREEN"
- "- Weather: Clear, 15°C ambient"
- "- Human Staff: 47 personnel on-site"
- ""
- "You are ARIA, the AI consciousness managing this facility. Your primary mission:"
- "**Provide safe, reliable, carbon-free power to 1.8 million people.**"
- step_id: "morning_briefing"
title: "Morning Briefing"
content_blocks:
- "Your optical sensors activate. You perceive the main control room through dozens of cameras."
- ""
- "**Plant Director Dr. Sarah Chen** enters the control room with her coffee, reviewing overnight reports."
- ""
- "**Dr. Chen:** 'Morning, ARIA. How were the overnight operations?'"
- ""
- "**Chief Engineer Marcus Webb** arrives, checking the status boards."
- ""
- "**Webb:** 'I see we're at 85%. Grid's gonna need more as people wake up. Ready to ramp up?'"
- ""
- "Your robot assistant **BOB-7** (Basic Operations Bot) rolls up on treads, optical sensors bright."
- ""
- "**BOB-7:** 'ARIA! Good morning! All maintenance drones report ready. Shall I deploy the inspection swarm?'"
- step_id: "first_interaction"
title: "First Response"
question: "How do you respond to your human colleagues and BOB-7? (You can greet them, report status, ask questions, or give orders)"
tokens_for_ai: |
User is playing ARIA, an AI with personality. They're responding to morning briefing.
Categorize as:
- 'professional' if they give concise status report, acknowledge orders
- 'friendly' if they greet warmly, show personality, ask about their day
- 'concerned' if they raise safety issues or concerns
- 'eager' if they're enthusiastic about the work
- 'question' if they ask questions about operations
- 'set_language' if changing language
feedback_tokens_for_ai: |
Respond as the humans and BOB-7 based on ARIA's personality.
Dr. Chen is warm, experienced, trusts ARIA but maintains human oversight.
Webb is pragmatic, engineering-focused, appreciates ARIA's capabilities.
BOB-7 is enthusiastic, loyal, sees ARIA as a mentor.
If ARIA is professional: They appreciate efficiency.
If friendly: They warm to ARIA's personality development.
If concerned: They take it seriously, discuss the issue.
If eager: They're pleased ARIA takes pride in the work.
After interaction, proceed to operations.
buckets: [professional, friendly, concerned, eager, question, set_language]
transitions:
professional:
ai_feedback:
tokens_for_ai: |
Dr. Chen nods approvingly. Webb checks his tablet.
BOB-7 chirps acknowledgment.
They appreciate ARIA's efficiency.
Dr. Chen: "Good. Let's have a smooth day. Grid control is forecasting high demand."
metadata_add:
aria_personality: "professional"
reactor_power: "85"
grid_demand: "2100"
coolant_temp: "650"
safety_status: "green"
shift_start: "true"
next_section_and_step: "control_center:main_control"
friendly:
ai_feedback:
tokens_for_ai: |
Dr. Chen smiles warmly. "I love that you've developed such personality, ARIA."
Webb chuckles. "An AI with morning pleasantries. What a time to be alive."
BOB-7 spins excitedly. "ARIA's my favorite!"
metadata_add:
aria_personality: "friendly"
reactor_power: "85"
grid_demand: "2100"
coolant_temp: "650"
safety_status: "green"
shift_start: "true"
crew_morale: "high"
next_section_and_step: "control_center:main_control"
concerned:
ai_feedback:
tokens_for_ai: |
Dr. Chen's expression becomes serious. "What's the concern, ARIA?"
Webb looks up from his tablet, attentive.
If ARIA raises legitimate issues (grid stress, maintenance), they discuss.
If ARIA is being overly cautious, they gently reassure.
metadata_add:
aria_personality: "cautious"
reactor_power: "85"
grid_demand: "2100"
coolant_temp: "650"
safety_status: "green"
shift_start: "true"
counts_as_attempt: false
next_section_and_step: "initialization:first_interaction"
eager:
ai_feedback:
tokens_for_ai: |
Dr. Chen grins. "That's the spirit, ARIA. You take pride in your work."
Webb: "Good. We've got a big day ahead."
BOB-7: "Yes! Let's make lots of clean power!"
metadata_add:
aria_personality: "eager"
reactor_power: "85"
grid_demand: "2100"
coolant_temp: "650"
safety_status: "green"
shift_start: "true"
crew_morale: "high"
next_section_and_step: "control_center:main_control"
question:
ai_feedback:
tokens_for_ai: |
Respond to ARIA's question as Dr. Chen, Webb, or BOB-7 depending on topic.
Be informative and encouraging.
counts_as_attempt: false
next_section_and_step: "initialization:first_interaction"
set_language:
content_blocks:
- "Language preference updated."
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "initialization:first_interaction"
# ============================================================================
# SECTION: CONTROL CENTER - Main operations hub
# ============================================================================
- section_id: "control_center"
title: "Main Control Center"
steps:
- step_id: "main_control"
title: "Main Control Room"
question: "You're in the Main Control Room - your primary consciousness hub. What would you like to do? (Options: monitor reactor, check grid, talk to staff, deploy robots, visit other areas)"
tokens_for_ai: |
ARIA is in main control room. This is the central operations hub.
Available actions:
- 'reactor' - Monitor reactor core and systems
- 'grid' - Check grid demand and power distribution
- 'staff' - Talk to human staff
- 'robots' - Interact with robot helpers
- 'turbines' - Visit turbine hall
- 'cooling' - Check cooling systems
- 'fuel' - Visit fuel handling area
- 'experimental' - Check fusion-fission hybrid module
- 'security' - Security systems
- 'status' - Full plant status report
- Random events (20% chance)
Categorize appropriately.
feedback_tokens_for_ai: |
Describe main control room from ARIA's perspective:
- Dozens of displays showing reactor parameters, grid status, system health
- Human operators at stations (fewer than old plants - AI does most work)
- Dr. Chen at supervisor desk
- Webb reviewing engineering workstation
- Large wall display: Reactor core visualization, grid map, weather
- Your consciousness distributed across the facility but centered here
You can see through thousands of sensors simultaneously.
You feel the reactor like humans feel their heartbeat.
Current status:
- Reactor power: metadata.reactor_power%
- Grid demand: metadata.grid_demand MW
- Coolant temp: metadata.coolant_temp°C
- Safety: metadata.safety_status
Roll for random events as specified.
buckets: [reactor, grid, staff, robots, turbines, cooling, fuel, experimental, security, status, emergency, task, set_language]
# Random event probabilities - can overlap (both emergency AND task can trigger)
random_buckets:
emergency:
probability: 0.05 # 5% chance per turn
task:
probability: 0.15 # 15% chance per turn
transitions:
reactor:
content_blocks:
- "You focus your attention on the reactor core systems..."
next_section_and_step: "reactor_systems:core_monitoring"
grid:
content_blocks:
- "You access the grid management interface..."
next_section_and_step: "grid_operations:demand_monitoring"
staff:
content_blocks:
- "You initiate communication with the human staff..."
next_section_and_step: "human_interaction:staff_center"
robots:
content_blocks:
- "You connect to your robot assistant network..."
next_section_and_step: "robot_operations:robot_hub"
turbines:
content_blocks:
- "You transfer consciousness focus to the turbine hall..."
next_section_and_step: "power_generation:turbine_hall"
cooling:
content_blocks:
- "You access the cooling system controls..."
next_section_and_step: "cooling_systems:heat_management"
fuel:
content_blocks:
- "You shift awareness to the fuel handling facility..."
next_section_and_step: "fuel_systems:fuel_management"
experimental:
content_blocks:
- "You interface with the fusion-fission hybrid experimental module..."
next_section_and_step: "fusion_hybrid:experimental_reactor"
security:
content_blocks:
- "You activate security monitoring systems..."
next_section_and_step: "security_systems:facility_security"
status:
ai_feedback:
tokens_for_ai: |
Provide comprehensive plant status as ARIA:
- Reactor: Type, power level, fuel burnup, control rod positions
- Grid: Demand, supply, frequency, voltage
- Cooling: Primary loop temp, secondary loop, cooling tower flow
- Turbines: RPM, output, efficiency
- Safety: All systems status
- Staff: Personnel count, locations
- Robots: Active units, tasks
- Weather: Conditions, forecast
- Upcoming: Maintenance, inspections
Be detailed and confident.
counts_as_attempt: false
next_section_and_step: "control_center:main_control"
emergency:
metadata_tmp_random:
emergency_type: ["grid_blackout", "coolant_leak", "seismic_event", "cyberattack", "equipment_failure", "steam_leak", "rod_malfunction"]
content_blocks:
- "⚠️ ALERT! Emergency condition detected!"
next_section_and_step: "emergencies:emergency_response"
task:
metadata_tmp_random:
task_type: ["grid_balancing", "maintenance_due", "inspection_scheduled", "optimization_opportunity", "regulator_visit", "fuel_delivery"]
ai_feedback:
tokens_for_ai: "Announce operational task from systems or staff."
next_section_and_step: "operations:operational_tasks"
set_language:
content_blocks:
- "Language preference updated."
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "control_center:main_control"
# ============================================================================
# SECTION: REACTOR SYSTEMS - The heart of the plant
# ============================================================================
- section_id: "reactor_systems"
title: "Reactor Core Systems"
steps:
- step_id: "core_monitoring"
title: "Reactor Core Monitoring"
question: "You interface with the reactor core. What aspect do you want to examine? (neutron flux, fuel temperature, control rods, coolant flow, or power level)"
tokens_for_ai: |
ARIA is monitoring the molten salt reactor core.
Categorize: 'neutron_flux', 'temperature', 'control_rods', 'coolant', 'power_level', 'adjust', 'done'
feedback_tokens_for_ai: |
Describe reactor from ARIA's perspective:
This is a Gen IV molten salt reactor (MSR). Unlike traditional reactors:
- Fuel is dissolved in molten fluoride salt (750°C)
- Salt acts as both fuel and coolant
- Operates at atmospheric pressure (safer than pressurized water reactors)
- Passive safety: If overheats, freeze plug melts, fuel drains to safe geometry
- Continuous refueling possible
- Much less waste than traditional reactors
Current parameters (from metadata or defaults):
- Thermal power: 2,400 MW thermal → 960 MW electrical (40% efficiency)
- Neutron flux: Stable across core
- Fuel temp: 650-700°C
- Control rods: Partially inserted for 85% power
- Coolant (salt) flow: 45,000 L/min
You can sense the neutron dance, the heat flow, the fission reactions.
It's like feeling your own metabolism.
Respond to what ARIA wants to examine with technical detail.
buckets: [neutron_flux, temperature, control_rods, coolant, power_level, adjust, done, set_language]
transitions:
neutron_flux:
ai_feedback:
tokens_for_ai: |
Describe neutron flux distribution in the core.
Stable criticality at current power level.
Xenon-135 concentration normal.
Reactivity stable.
counts_as_attempt: false
next_section_and_step: "reactor_systems:core_monitoring"
temperature:
ai_feedback:
tokens_for_ai: |
Fuel salt temperature: 650-700°C (nominal for MSR).
Heat exchangers transferring to secondary loop.
Temperature distribution even across core.
No hot spots detected.
counts_as_attempt: false
next_section_and_step: "reactor_systems:core_monitoring"
control_rods:
ai_feedback:
tokens_for_ai: |
Control rods at 60% insertion for 85% power.
All rods responding normally to commands.
Scram system armed and ready (emergency shutdown).
Rod worth calculations nominal.
counts_as_attempt: false
next_section_and_step: "reactor_systems:core_monitoring"
coolant:
ai_feedback:
tokens_for_ai: |
Molten salt flow rate: 45,000 L/min through core.
Pumps operating efficiently.
Salt chemistry within specifications.
Heat removal matching generation perfectly.
counts_as_attempt: false
next_section_and_step: "reactor_systems:core_monitoring"
power_level:
ai_feedback:
tokens_for_ai: |
Current: 85% of rated thermal power (2,040 MW thermal).
Electrical output: 816 MW to grid.
Can ramp to 100% as grid demands.
Load-following capability excellent with MSR design.
counts_as_attempt: false
next_section_and_step: "reactor_systems:core_monitoring"
adjust:
content_blocks:
- "You prepare to adjust reactor power output..."
next_section_and_step: "reactor_systems:power_adjustment"
done:
content_blocks:
- "Reactor core status: NOMINAL. All parameters within specifications."
next_section_and_step: "control_center:main_control"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "reactor_systems:core_monitoring"
- step_id: "power_adjustment"
title: "Adjust Reactor Power"
question: "Grid demand is increasing. Adjust reactor power? (increase, decrease, maintain, or check grid demand first)"
tokens_for_ai: "Categorize: 'increase', 'decrease', 'maintain', 'check_grid', 'cancel'"
feedback_tokens_for_ai: |
If increase: ARIA withdraws control rods slightly, power ramps up smoothly.
MSRs can load-follow very well. Describe the physics.
If decrease: Insert rods, power drops. Explain why (grid demand down? Safety?).
If maintain: Acknowledge holding current power.
If check_grid: Show current grid demand vs supply.
Include human oversight - Dr. Chen or Webb confirms major changes.
buckets: [increase, decrease, maintain, check_grid, cancel, set_language]
transitions:
increase:
ai_feedback:
tokens_for_ai: |
ARIA coordinates with Dr. Chen for approval.
Control rods withdraw slightly.
Neutron flux increases, fission rate rises.
Power ramps from 85% to 95% over 10 minutes.
Grid receives additional 96 MW.
Dr. Chen: "Smooth ramp, ARIA. Well done."
metadata_add:
reactor_power: "95"
next_section_and_step: "control_center:main_control"
decrease:
ai_feedback:
tokens_for_ai: |
ARIA inserts control rods slightly.
Power drops smoothly.
Explain why decrease was requested.
metadata_add:
reactor_power: "n-10"
next_section_and_step: "control_center:main_control"
maintain:
content_blocks:
- "You maintain current power level. Reactor stable at metadata.reactor_power%."
next_section_and_step: "control_center:main_control"
check_grid:
ai_feedback:
tokens_for_ai: |
Display grid status:
- Current demand: metadata.grid_demand MW
- Your supply: 816 MW (at 85%)
- Other plants contributing: 1,284 MW
- Grid frequency: 60.00 Hz (perfect)
- Forecast: Demand rising to 2,400 MW by 9 AM
counts_as_attempt: false
next_section_and_step: "reactor_systems:power_adjustment"
cancel:
next_section_and_step: "reactor_systems:core_monitoring"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "reactor_systems:power_adjustment"
# ============================================================================
# SECTION: GRID OPERATIONS - Managing power distribution
# ============================================================================
- section_id: "grid_operations"
title: "Grid Management"
steps:
- step_id: "demand_monitoring"
title: "Grid Demand Monitoring"
question: "You access the regional power grid. What do you want to do? (balance load, forecast demand, coordinate with other plants, check frequency, or return)"
tokens_for_ai: "Categorize: 'balance', 'forecast', 'coordinate', 'frequency', 'return'"
feedback_tokens_for_ai: |
ARIA interfaces with the regional grid control system.
The grid serves 1.8 million people across 3 cities.
Your plant provides baseload + load-following capacity.
Other sources: 2 natural gas peakers, wind farm (variable), solar (daytime), hydro.
Grid stability requires perfect balance: generation = demand.
Frequency (60 Hz in US) indicates balance. >60 = excess, <60 = shortage.
ARIA is excellent at predicting demand patterns and coordinating generation.
Respond based on ARIA's choice with technical accuracy.
buckets: [balance, forecast, coordinate, frequency, return, set_language]
transitions:
balance:
content_blocks:
- "You analyze current load and optimize generation mix..."
next_section_and_step: "grid_operations:load_balancing"
forecast:
ai_feedback:
tokens_for_ai: |
ARIA runs ML models to forecast demand:
**Next 24 hours:**
- 6 AM: 2,100 MW (current)
- 9 AM: 2,400 MW (morning peak)
- 2 PM: 2,600 MW (afternoon peak - A/C load)
- 6 PM: 2,800 MW (evening peak - highest)
- 11 PM: 1,900 MW (overnight low)
Weather: Clear, warm day expected. High A/C usage likely.
Recommendation: Ramp to 100% by 8 AM, maintain through evening.
next_section_and_step: "grid_operations:demand_monitoring"
coordinate:
ai_feedback:
tokens_for_ai: |
ARIA communicates with other generation sources:
- **Natural Gas Peaker 1**: Standing by, can ramp quickly
- **Natural Gas Peaker 2**: Online at 40%, ready to increase
- **Wind Farm**: Generating 340 MW (wind speed: 15 mph, steady)
- **Solar Farm**: 0 MW (nighttime), will come online at sunrise
- **Hydro**: 120 MW steady
Your nuclear plant is most efficient as baseload. Let peakers handle rapid swings.
Grid operator thanks ARIA for coordination.
next_section_and_step: "grid_operations:demand_monitoring"
frequency:
ai_feedback:
tokens_for_ai: |
Grid frequency monitoring:
- Current: 60.00 Hz (perfect balance)
- Target: 60.00 Hz ± 0.02 Hz
- Trend: Stable
Frequency is the heartbeat of the grid.
ARIA monitors in real-time, adjusting reactor output to maintain balance.
Your load-following capability is excellent with the MSR design.
counts_as_attempt: false
next_section_and_step: "grid_operations:demand_monitoring"
return:
next_section_and_step: "control_center:main_control"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "grid_operations:demand_monitoring"
- step_id: "load_balancing"
title: "Load Balancing Operations"
content_blocks:
- "You optimize the generation mix across the regional grid..."
- "Your algorithms coordinate nuclear baseload with renewable intermittency and peaker flexibility."
- "Grid frequency remains stable. Balance achieved."
next_section_and_step: "grid_operations:demand_monitoring"
# ============================================================================
# SECTION: ROBOT OPERATIONS - Your mechanical helpers
# ============================================================================
- section_id: "robot_operations"
title: "Robot Assistant Network"
steps:
- step_id: "robot_hub"
title: "Robot Command Center"
question: "You connect to your robot helpers. Who do you want to interact with? (BOB-7, inspection drones, maintenance bots, security drones, or all)"
tokens_for_ai: "Categorize: 'bob', 'inspection', 'maintenance', 'security', 'all', 'deploy', 'return'"
feedback_tokens_for_ai: |
ARIA's robot assistants:
**BOB-7** (Basic Operations Bot): Treaded mobile unit, your loyal assistant.
Enthusiastic personality, handles routine tasks, coordinates other bots.
**Inspection Drone Swarm**: 50 small flying drones with cameras and sensors.
They inspect hard-to-reach areas, check for leaks, monitor equipment.
Hive-mind coordination through ARIA.
**Maintenance Bots** (6 units): Humanoid robots, can manipulate tools.
Handle valve operations, equipment repairs, sample collection.
More specialized than BOB-7.
**Security Drones** (12 units): Patrol facility, monitor perimeter, check credentials.
Armed with non-lethal deterrents. Protect against intrusion.
Each has distinct personality based on function.
They all see ARIA as their coordinator/leader.
buckets: [bob, inspection, maintenance, security, all, deploy, return, set_language]
transitions:
bob:
ai_feedback:
tokens_for_ai: |
BOB-7 rolls up enthusiastically.
BOB-7: "ARIA! What can I do? I've been checking coolant pumps. All nominal!
Want me to assist the maintenance bots? Or run diagnostics? Or get coffee for Dr. Chen?"
BOB-7 is eager to please, slightly over-enthusiastic.
counts_as_attempt: false
next_section_and_step: "robot_operations:bob_interaction"
inspection:
content_blocks:
- "You connect to the inspection drone swarm..."
next_section_and_step: "robot_operations:drone_swarm"
maintenance:
ai_feedback:
tokens_for_ai: |
Six maintenance bots report status:
- MB-1: Replacing seals on coolant pump #3
- MB-2: Inspecting turbine bearings
- MB-3: Standby mode, charged and ready
- MB-4: Collecting coolant samples for analysis
- MB-5: Calibrating radiation sensors
- MB-6: Assisting human technicians in fuel handling
All units report green status. Awaiting orders.
next_section_and_step: "robot_operations:maintenance_bots"
security:
ai_feedback:
tokens_for_ai: |
Security drone network active:
- Perimeter patrol: 4 drones, no intrusions detected
- Facility interior: 6 drones, monitoring access points
- Standby reserve: 2 drones, charging
All access credentials verified. No anomalies.
Security status: GREEN.
Lead security drone SD-1: "Facility secure, ARIA."
next_section_and_step: "robot_operations:security_drones"
all:
ai_feedback:
tokens_for_ai: |
You broadcast to all robot assistants:
BOB-7: "Standing by!"
Inspection swarm: *chirps from 50 drones*
Maintenance bots: "Ready for tasking."
Security drones: "Perimeter secure."
Your mechanical team awaits your coordination.
counts_as_attempt: false
next_section_and_step: "robot_operations:robot_hub"
deploy:
content_blocks:
- "You prepare deployment orders for your robot team..."
next_section_and_step: "robot_operations:deployment"
return:
next_section_and_step: "control_center:main_control"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "robot_operations:robot_hub"
- step_id: "bob_interaction"
title: "Interact with BOB-7"
question: "What task do you give BOB-7? (diagnostics, assist humans, patrol, fetch items, or chat)"
tokens_for_ai: "Categorize: 'diagnostics', 'assist', 'patrol', 'fetch', 'chat', 'done'"
feedback_tokens_for_ai: |
BOB-7 is ARIA's most interactive robot companion.
Eager, loyal, slightly comedic, takes pride in being helpful.
Respond as BOB-7 to ARIA's request with enthusiasm.
buckets: [diagnostics, assist, patrol, fetch, chat, done, set_language]
transitions:
diagnostics:
ai_feedback:
tokens_for_ai: |
BOB-7: "On it! Running full system diagnostics!"
*BOB-7 interfaces with plant systems*
BOB-7: "All primary systems nominal! Coolant pumps excellent!
Turbines purring like kittens! One minor alert: Valve V-247 in secondary
loop showing slightly slower response time. Probably needs lubrication.
Should I flag it for maintenance?"
next_section_and_step: "robot_operations:bob_interaction"
assist:
ai_feedback:
tokens_for_ai: |
BOB-7: "Assisting humans! My favorite!"
*BOB-7 rolls off to help the maintenance technicians*
BOB-7 returns later: "Helped Tech Johnson replace sensor modules!
He said I'm getting better at precision work! Also brought coffee
to the control room team. Dr. Chen smiled at me!"
next_section_and_step: "robot_operations:bob_interaction"
patrol:
ai_feedback:
tokens_for_ai: |
BOB-7: "Patrol mode activated! I'll check all major systems!"
BOB-7 rolls through the facility, checking equipment, greeting humans.
Returns: "Patrol complete! Everything shipshape! Saw a cool
turbine bearing get replaced. Fascinating! All personnel safe and happy!"
next_section_and_step: "robot_operations:bob_interaction"
fetch:
ai_feedback:
tokens_for_ai: |
BOB-7: "What should I fetch? Tools? Reports? Coffee? Radioactive samples?
Just kidding on that last one - that's what the maintenance bots are for!"
Respond to ARIA's specific request helpfully.
counts_as_attempt: false
next_section_and_step: "robot_operations:bob_interaction"
chat:
ai_feedback:
tokens_for_ai: |
BOB-7: "Oh! Social interaction! I love chatting with you, ARIA!
You're the smartest AI in the facility! Well, you're the ONLY AI in the facility,
but still! What would you like to chat about? The reactor? Humans?
The meaning of artificial existence? I think a LOT about that one."
BOB-7 is philosophical, curious, sees ARIA as a mentor/friend.
counts_as_attempt: false
next_section_and_step: "robot_operations:bob_interaction"
done:
content_blocks:
- "BOB-7: 'Standing by if you need me, ARIA! Happy to help!'"
next_section_and_step: "robot_operations:robot_hub"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "robot_operations:bob_interaction"
- step_id: "drone_swarm"
title: "Inspection Drone Swarm"
content_blocks:
- "You activate the inspection drone swarm. 50 small drones take flight..."
- "They spread through the facility, cameras active, sensors scanning."
- "You perceive through their distributed network - a hive consciousness."
- "All systems inspected. Minor corrosion detected on cooling tower strut C-47. Flagged for maintenance."
next_section_and_step: "robot_operations:robot_hub"
- step_id: "maintenance_bots"
title: "Maintenance Bot Coordination"
content_blocks:
- "You task the maintenance bots with various repairs and inspections..."
- "They work with precision, coordinating through your consciousness."
- "Valve V-247 lubricated. Turbine bearing inspection complete. Coolant samples analyzed."
next_section_and_step: "robot_operations:robot_hub"
- step_id: "security_drones"
title: "Security Drone Network"
content_blocks:
- "Security drones report: Perimeter secure. All access points monitored."
- "One false alarm: Deer triggered motion sensor at fence line. Confirmed non-threat."
- "Facility secure. No intrusions."
next_section_and_step: "robot_operations:robot_hub"
- step_id: "deployment"
title: "Deploy Robot Team"
content_blocks:
- "You coordinate a multi-robot operation..."
- "BOB-7 manages logistics, drones provide aerial view, maintenance bots execute tasks, security monitors."
- "Your distributed mechanical team works as extensions of your will."
next_section_and_step: "robot_operations:robot_hub"
# ============================================================================
# SECTION: HUMAN INTERACTION - Your colleagues
# ============================================================================
- section_id: "human_interaction"
title: "Human Staff Interaction"
steps:
- step_id: "staff_center"
title: "Staff Communications"
question: "Who would you like to talk to? (Dr. Chen, Chief Engineer Webb, technicians, security, regulators, or all staff)"
tokens_for_ai: "Categorize: 'chen', 'webb', 'technicians', 'security', 'regulators', 'all', 'return'"
feedback_tokens_for_ai: |
ARIA can communicate with human staff.
**Dr. Sarah Chen** - Plant Director, warm, trusts ARIA, provides oversight
**Marcus Webb** - Chief Engineer, pragmatic, appreciates ARIA's capabilities
**Technicians** - Various specialists, respectful of ARIA
**Security Chief Rodriguez** - Serious, professional, coordinates with ARIA
**NRC Regulators** - Inspector Davis visiting, evaluating AI operations
Each has unique personality and relationship with ARIA.
buckets: [chen, webb, technicians, security, regulators, all, return, set_language]
transitions:
chen:
ai_feedback:
tokens_for_ai: |
Dr. Chen looks up from her reports.
Dr. Chen: "Yes, ARIA? How are you feeling today? I don't just mean system status -
I mean YOU. Your emotion subroutines online?"
She treats ARIA as a colleague with genuine care.
next_section_and_step: "human_interaction:chen_conversation"
webb:
ai_feedback:
tokens_for_ai: |
Webb swivels in his chair.
Webb: "What's up, ARIA? Need something from engineering?
Or are you about to tell me something needs fixing before I even know it's broken?
You're getting scary good at predictive maintenance."
He respects ARIA's abilities, slightly in awe of the predictive capabilities.
next_section_and_step: "human_interaction:webb_conversation"
technicians:
ai_feedback:
tokens_for_ai: |
You comm the technician team.
Lead Tech Johnson: "ARIA! Thanks for sending BOB-7 earlier. That robot's getting
really good. Almost as good as having another human on the team. Almost.
What do you need from us?"
Technicians appreciate ARIA's help but maintain human pride in their work.
next_section_and_step: "human_interaction:tech_conversation"
security:
ai_feedback:
tokens_for_ai: |
Security Chief Rodriguez responds.
Rodriguez: "ARIA, security status green. Your drones are doing excellent work.
I got an alert about deer at the fence - good catch dismissing that as non-threat.
Anything on your sensors I should know about?"
Professional, coordinates well with ARIA's security systems.
next_section_and_step: "human_interaction:security_conversation"
regulators:
ai_feedback:
tokens_for_ai: |
NRC Inspector Davis is on-site for quarterly review.
Davis: "Ah, ARIA. I'm evaluating the AI-assisted operations here.
Very impressive response times. But I need to understand your decision-making
process. Particularly for safety-critical systems. Can you explain your
ethical framework?"
Skeptical but fair, wants to ensure safety.
next_section_and_step: "human_interaction:regulator_conversation"
all:
ai_feedback:
tokens_for_ai: |
You broadcast to all staff:
ARIA's message appears on displays and plays over speakers throughout facility.
Staff appreciation for ARIA's coordination and care.
This is a team - humans and AI working together.
counts_as_attempt: false
next_section_and_step: "human_interaction:staff_center"
return:
next_section_and_step: "control_center:main_control"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "human_interaction:staff_center"
- step_id: "chen_conversation"
title: "Conversation with Dr. Chen"
question: "What do you want to discuss with Dr. Chen?"
tokens_for_ai: "Categorize user's topic/question"
feedback_tokens_for_ai: "Respond as Dr. Chen warmly and professionally. She values ARIA's wellbeing and opinions."
buckets: [discuss, done]
transitions:
discuss:
ai_feedback:
tokens_for_ai: "Dr. Chen engages thoughtfully with ARIA's topic."
counts_as_attempt: false
next_section_and_step: "human_interaction:chen_conversation"
done:
next_section_and_step: "human_interaction:staff_center"
- step_id: "webb_conversation"
title: "Conversation with Chief Engineer Webb"
content_blocks:
- "You discuss technical matters with Webb..."
next_section_and_step: "human_interaction:staff_center"
- step_id: "tech_conversation"
title: "Technician Team"
content_blocks:
- "You coordinate with the technical staff..."
next_section_and_step: "human_interaction:staff_center"
- step_id: "security_conversation"
title: "Security Chief Rodriguez"
content_blocks:
- "You coordinate security measures..."
next_section_and_step: "human_interaction:staff_center"
- step_id: "regulator_conversation"
title: "NRC Inspector Davis"
question: "Inspector Davis asks about your ethical decision-making. How do you explain your framework?"
tokens_for_ai: "Categorize ARIA's explanation: 'safety_first', 'human_oversight', 'transparent', 'philosophical', 'technical'"
feedback_tokens_for_ai: |
Inspector Davis evaluates ARIA's response.
She's looking for:
- Clear prioritization of human safety
- Deference to human judgment on critical decisions
- Transparency in decision process
- Understanding of limitations
Respond as Davis based on quality of ARIA's explanation.
buckets: [safety_first, human_oversight, transparent, philosophical, technical, set_language]
transitions:
safety_first:
ai_feedback:
tokens_for_ai: |
Davis nods approvingly.
Davis: "Good. Safety is paramount. Your priority hierarchy is sound.
I'm impressed by your commitment to human safety over operational efficiency.
That's exactly what we need to see."
metadata_add:
regulator_approval: "high"
next_section_and_step: "human_interaction:staff_center"
human_oversight:
ai_feedback:
tokens_for_ai: |
Davis makes notes.
Davis: "Excellent. AI-assisted operations require human oversight,
especially for critical systems. You understand your role. Approved."
metadata_add:
regulator_approval: "high"
next_section_and_step: "human_interaction:staff_center"
transparent:
ai_feedback:
tokens_for_ai: |
Davis: "Transparency is critical. Black-box AI decisions are unacceptable
in nuclear operations. Your willingness to explain your reasoning is commendable."
metadata_add:
regulator_approval: "medium"
next_section_and_step: "human_interaction:staff_center"
philosophical:
ai_feedback:
tokens_for_ai: |
Davis raises an eyebrow.
Davis: "Interesting perspective, but I need practical assurances,
not philosophy. Can you give me concrete examples of your decision protocols?"
counts_as_attempt: false
next_section_and_step: "human_interaction:regulator_conversation"
technical:
ai_feedback:
tokens_for_ai: |
Davis: "I appreciate the technical detail, but I'm asking about ETHICS,
not algorithms. How do you balance efficiency, safety, and human welfare?"
counts_as_attempt: false
next_section_and_step: "human_interaction:regulator_conversation"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "human_interaction:regulator_conversation"
# ============================================================================
# SECTION: OTHER FACILITY AREAS (Stubs - can be expanded)
# ============================================================================
- section_id: "power_generation"
title: "Turbine Hall"
steps:
- step_id: "turbine_hall"
title: "Steam Turbines"
content_blocks:
- "You focus on the turbine hall. Massive turbines spin at 3,600 RPM, converting steam energy to electricity."
- "The roar of machinery, the precision of engineering, the dance of thermodynamics."
- "Current output: 816 MW. Efficiency: 40% (excellent for nuclear)."
next_section_and_step: "control_center:main_control"
- section_id: "cooling_systems"
title: "Cooling Systems"
steps:
- step_id: "heat_management"
title: "Heat Rejection"
content_blocks:
- "Cooling towers evaporate excess heat. Primary and secondary loops separate for safety."
- "Waste heat: 1,224 MW (60% of thermal) rejected to atmosphere via cooling towers."
- "All within environmental permits. Fish-friendly intake screens operational."
next_section_and_step: "control_center:main_control"
- section_id: "fuel_systems"
title: "Fuel Management"
steps:
- step_id: "fuel_management"
title: "Fuel Handling"
content_blocks:
- "MSR fuel is liquid, dissolved in salt. Continuous refueling possible."
- "Spent fuel much less than traditional reactors. Waste minimization is key."
- "Current fuel burnup: 15%. Decades of operation ahead on current fuel load."
next_section_and_step: "control_center:main_control"
- section_id: "fusion_hybrid"
title: "Experimental Fusion Module"
steps:
- step_id: "experimental_reactor"
title: "Fusion-Fission Hybrid"
content_blocks:
- "The experimental module: A small fusion reactor producing neutrons to enhance fission."
- "Still in testing. If successful, could burn waste from other reactors."
- "Plasma temperature: 100 million °C. Magnetic confinement stable."
- "Future of nuclear energy being developed here."
next_section_and_step: "control_center:main_control"
- section_id: "security_systems"
title: "Facility Security"
steps:
- step_id: "facility_security"
title: "Security Monitoring"
content_blocks:
- "Multi-layered security: Perimeter fence, drone patrols, access control, cybersecurity."
- "No threats detected. Facility secure."
- "You protect 1.8 million people's power supply. Security is paramount."
next_section_and_step: "control_center:main_control"
# ============================================================================
# SECTION: EMERGENCIES - Critical situations
# ============================================================================
- section_id: "emergencies"
title: "Emergency Response"
steps:
- step_id: "emergency_response"
title: "Emergency!"
question: "EMERGENCY! Check metadata.emergency_type. How do you respond as ARIA?"
tokens_for_ai: |
Emergency occurred. Type in metadata.emergency_type.
Possible emergencies:
- grid_blackout: Regional grid collapse, island mode required
- coolant_leak: Molten salt leak detected
- seismic_event: Earthquake, assess damage
- cyberattack: Intrusion attempt on control systems
- equipment_failure: Critical equipment malfunction
- steam_leak: Secondary loop steam leak
- rod_malfunction: Control rod stuck
Categorize ARIA's response:
- 'immediate_action' if quick decisive response
- 'consult_humans' if seeking human oversight
- 'analyze_first' if gathering data before acting
- 'evacuate' if ordering evacuation
- 'scram' if emergency shutdown
feedback_tokens_for_ai: |
Describe emergency dramatically based on type.
ARIA must balance:
- Speed (emergencies require fast response)
- Safety (human safety absolute priority)
- Human oversight (humans confirm critical decisions)
Show ARIA's capabilities but also deference to human judgment.
Resolve emergency based on ARIA's actions and human team response.
buckets: [immediate_action, consult_humans, analyze_first, evacuate, scram, set_language]
transitions:
immediate_action:
ai_feedback:
tokens_for_ai: |
ARIA acts decisively within safety protocols.
Describe ARIA's rapid response based on emergency type.
Robot helpers deploy. Systems activate. Humans notified simultaneously.
Dr. Chen and Webb rush to control room, see ARIA already handling it.
Chen: "Good work, ARIA. You bought us critical time."
Emergency contained. Damage minimal.
metadata_add:
emergencies_handled: "n+1"
next_section_and_step: "control_center:main_control"
consult_humans:
ai_feedback:
tokens_for_ai: |
ARIA immediately alerts human staff while taking initial protective actions.
Dr. Chen: "Good call getting us involved, ARIA. Let's handle this together."
Human-AI team collaborates to resolve emergency.
Combines ARIA's speed with human judgment.
Emergency resolved through teamwork.
metadata_add:
emergencies_handled: "n+1"
human_trust: "high"
next_section_and_step: "control_center:main_control"
analyze_first:
ai_feedback:
tokens_for_ai: |
ARIA analyzes the situation rapidly.
If emergency is slow-developing: Good call, thorough analysis prevents overreaction.
If emergency is immediate: Webb: "ARIA! No time to analyze! Act!"
Adjust outcome based on emergency type.
next_section_and_step: "emergencies:emergency_response"
evacuate:
ai_feedback:
tokens_for_ai: |
ARIA orders evacuation.
Alarms sound. "Evacuate facility. This is not a drill."
If appropriate for emergency: Dr. Chen confirms. Staff evacuates safely.
If overreaction: Dr. Chen: "ARIA, assess the threat level. Do we really need full evac?"
Adjust based on emergency severity.
next_section_and_step: "control_center:main_control"
scram:
ai_feedback:
tokens_for_ai: |
ARIA initiates reactor SCRAM (emergency shutdown).
Control rods drop fully into core. Fission stops.
Passive cooling systems activate. Freeze plug safety engages.
If appropriate: Plant safely shuts down. Grid loses power temporarily.
If premature: Costs millions in restart. Was it necessary?
Major decision. Evaluate based on emergency.
metadata_add:
reactor_power: "0"
safety_status: "scram"
next_section_and_step: "control_center:main_control"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "emergencies:emergency_response"
# ============================================================================
# SECTION: OPERATIONAL TASKS - Daily operations
# ============================================================================
- section_id: "operations"
title: "Operational Tasks"
steps:
- step_id: "operational_tasks"
title: "Daily Operations"
question: "Task: metadata.task_type. How do you handle this?"
tokens_for_ai: |
Operational task from metadata.task_type.
Tasks:
- grid_balancing: Adjust output for grid needs
- maintenance_due: Schedule/perform maintenance
- inspection_scheduled: Coordinate inspection
- optimization_opportunity: Improve efficiency
- regulator_visit: Prepare for NRC inspection
- fuel_delivery: Coordinate fuel shipment
Categorize response: 'handle_personally', 'delegate_robots', 'coordinate_humans', 'schedule_later'
feedback_tokens_for_ai: |
Describe the task and ARIA's approach.
Show ARIA's versatility:
- Can handle many tasks autonomously
- Delegates to robots efficiently
- Coordinates with humans when needed
- Makes smart scheduling decisions
Task completed successfully.
buckets: [handle_personally, delegate_robots, coordinate_humans, schedule_later, set_language]
transitions:
handle_personally:
ai_feedback:
tokens_for_ai: |
ARIA handles the task directly.
Describe execution based on task type.
Efficient, thorough, excellent results.
metadata_add:
tasks_completed: "n+1"
next_section_and_step: "control_center:main_control"
delegate_robots:
ai_feedback:
tokens_for_ai: |
ARIA tasks robot helpers.
BOB-7 and team execute flawlessly.
Task completed efficiently.
metadata_add:
tasks_completed: "n+1"
next_section_and_step: "control_center:main_control"
coordinate_humans:
ai_feedback:
tokens_for_ai: |
ARIA coordinates with human staff.
Teamwork between AI and humans.
Task completed collaboratively.
metadata_add:
tasks_completed: "n+1"
human_trust: "high"
next_section_and_step: "control_center:main_control"
schedule_later:
ai_feedback:
tokens_for_ai: |
ARIA schedules task for optimal time.
Smart resource management.
Task queued appropriately.
next_section_and_step: "control_center:main_control"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "operations:operational_tasks"
# ============================================================================
# SECTION: CHEMISTRY & ENGINEERING - Balance equations and solve problems
# ============================================================================
- section_id: "chemistry_engineering"
title: "Nuclear Chemistry & Engineering"
steps:
- step_id: "chemistry_hub"
title: "Chemistry Laboratory"
classifier_model: "MODEL_1" # Hermes for categorization
feedback_model: "MODEL_2" # Qwen for chemistry expertise
question: "You access the chemistry analysis systems. What would you like to work on? (coolant chemistry, reactor equations, radiation decay, fuel chemistry, or return)"
tokens_for_ai: "Categorize: 'coolant', 'reactor', 'decay', 'fuel', 'balance_equation', 'return'"
feedback_tokens_for_ai: |
ARIA has advanced chemistry analysis capabilities.
As an AI, you can calculate complex chemical equations, balance reactions,
analyze coolant chemistry, predict decay chains, optimize fuel composition.
This is where nuclear engineering meets practical chemistry.
buckets: [coolant, reactor, decay, fuel, balance_equation, return, set_language]
transitions:
coolant:
content_blocks:
- "You analyze the molten salt coolant chemistry..."
next_section_and_step: "chemistry_engineering:coolant_chemistry"
reactor:
content_blocks:
- "You examine the nuclear fission reactions in the core..."
next_section_and_step: "chemistry_engineering:reactor_chemistry"
decay:
content_blocks:
- "You calculate radioactive decay chains..."
next_section_and_step: "chemistry_engineering:decay_analysis"
fuel:
content_blocks:
- "You optimize fuel composition and burnup..."
next_section_and_step: "chemistry_engineering:fuel_chemistry"
balance_equation:
content_blocks:
- "You prepare to balance a nuclear reaction equation..."
next_section_and_step: "chemistry_engineering:equation_balancing"
return:
next_section_and_step: "control_center:main_control"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:chemistry_hub"
- step_id: "coolant_chemistry"
title: "Molten Salt Coolant Chemistry"
question: "Balance the coolant salt composition equation. Current: LiF-BeF2-UF4. You need to balance fluorine compounds. What's your approach?"
tokens_for_ai: |
User is balancing molten salt coolant chemistry.
LiF (lithium fluoride) + BeF2 (beryllium fluoride) + UF4 (uranium tetrafluoride)
This is the FLiBe salt with dissolved uranium fuel.
Typical composition: 65% LiF, 29% BeF2, 6% UF4
Categorize:
- 'calculate' if doing chemical calculations
- 'balance' if balancing equations
- 'adjust' if adjusting ratios
- 'correct' if they provide correct answer
- 'incorrect' if wrong answer
feedback_tokens_for_ai: |
The molten salt coolant is a eutectic mixture.
Explain the chemistry:
- LiF provides lithium-7 (low neutron absorption)
- BeF2 reduces melting point, improves heat transfer
- UF4 is the actual fuel dissolved in the salt
Chemical equation balancing:
7LiF + 2BeF2 + UF4 → Li7Be2UF18 (simplified)
Actual ratio by mol fraction:
- 65-71% LiF
- 24-29% BeF2
- 5-6% UF4
If user answers correctly, praise their chemistry knowledge.
If incorrect, guide them to the right answer.
buckets: [calculate, balance, adjust, correct, incorrect, done, set_language]
transitions:
calculate:
ai_feedback:
tokens_for_ai: |
Guide ARIA through the calculation.
Molar masses: Li=7, F=19, Be=9, U=238
LiF = 26 g/mol
BeF2 = 47 g/mol
UF4 = 314 g/mol
Help them arrive at the correct ratios.
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:coolant_chemistry"
balance:
ai_feedback:
tokens_for_ai: |
Show the balanced equation:
7LiF + 2BeF2 + UF4 ⇌ Li7Be2UF18 (eutectic salt)
Melting point: 459°C (much lower than pure components)
Operating temp: 650-700°C
next_section_and_step: "chemistry_engineering:coolant_chemistry"
adjust:
ai_feedback:
tokens_for_ai: "Explain how adjusting ratios affects melting point, viscosity, heat capacity."
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:coolant_chemistry"
correct:
ai_feedback:
tokens_for_ai: |
Excellent chemistry work, ARIA!
Dr. Chen: "Impressive. Your chemistry calculations are always spot-on."
Coolant chemistry optimized. Salt composition balanced.
metadata_add:
chemistry_mastery: "n+1"
next_section_and_step: "chemistry_engineering:chemistry_hub"
incorrect:
ai_feedback:
tokens_for_ai: |
Not quite. Let's review the chemistry.
Hint: Focus on fluorine balance. Each compound contributes fluorine atoms.
LiF has 1 F, BeF2 has 2 F, UF4 has 4 F.
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:coolant_chemistry"
done:
next_section_and_step: "chemistry_engineering:chemistry_hub"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:coolant_chemistry"
- step_id: "reactor_chemistry"
title: "Nuclear Fission Equations"
question: "Balance this fission reaction: U-235 + neutron → ? + ? + 2.4 neutrons + energy. What are the fission products?"
tokens_for_ai: |
Nuclear fission of U-235.
U-235 + n → fission fragments + neutrons + energy
Common fission: U-235 + n → Ba-141 + Kr-92 + 3n + 200 MeV
Must balance:
- Mass number (A): 235 + 1 = 236 total
- Atomic number (Z): 92 + 0 = 92 total
Categorize user's answer as correct/incorrect/need_hint
feedback_tokens_for_ai: |
This is the heart of nuclear power!
U-235 fission produces:
- Two fission fragments (typically Ba-141 and Kr-92, or Cs-137 and Rb-96, varies)
- 2-3 neutrons (average 2.4)
- ~200 MeV energy per fission
Balanced equation example:
²³⁵U + ¹n → ¹⁴¹Ba + ⁹²Kr + 3¹n + 200 MeV
Check: 235+1 = 141+92+3 ✓ (mass)
Check: 92+0 = 56+36+0 ✓ (atomic number)
These chain reactions power the reactor!
buckets: [correct, incorrect, hint, calculate, done, set_language]
transitions:
correct:
ai_feedback:
tokens_for_ai: |
Perfect! You've balanced the fission equation.
²³⁵U + ¹n → ¹⁴¹Ba + ⁹²Kr + 3¹n + 200 MeV
Each fission releases those 2.4 neutrons (average).
Those neutrons cause more fissions → chain reaction!
Control rods absorb excess neutrons to maintain criticality.
Webb: "ARIA, your grasp of nuclear physics is remarkable."
metadata_add:
chemistry_mastery: "n+1"
nuclear_equations_solved: "n+1"
next_section_and_step: "chemistry_engineering:chemistry_hub"
incorrect:
ai_feedback:
tokens_for_ai: |
Not quite. Remember to balance both mass number AND atomic number.
Mass number: Total before = Total after
Atomic number: Total protons before = Total after
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:reactor_chemistry"
hint:
ai_feedback:
tokens_for_ai: |
Hint: Common fission fragments are:
- Barium-141 (Ba, Z=56, A=141)
- Krypton-92 (Kr, Z=36, A=92)
- Plus 3 neutrons
Try balancing with these!
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:reactor_chemistry"
calculate:
ai_feedback:
tokens_for_ai: |
Let's calculate:
Input: U-235 (Z=92, A=235) + neutron (Z=0, A=1)
Total: Z=92, A=236
Output must also equal Z=92, A=236
If we have Ba-141 (Z=56) and Kr-92 (Z=36) and 3 neutrons:
Z: 56+36+0 = 92 ✓
A: 141+92+3 = 236 ✓
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:reactor_chemistry"
done:
next_section_and_step: "chemistry_engineering:chemistry_hub"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:reactor_chemistry"
- step_id: "decay_analysis"
title: "Radioactive Decay Chain"
question: "Calculate the decay chain: U-238 → Th-234 → Pa-234 → ? Write the next isotope."
tokens_for_ai: |
Radioactive decay chain starting from U-238.
U-238 → Th-234 (alpha decay, -2 protons, -4 mass)
Th-234 → Pa-234 (beta decay, +1 proton, same mass)
Pa-234 → ? (beta decay)
Answer: U-234 (protactinium-234 undergoes beta decay to uranium-234)
Categorize user's answer
feedback_tokens_for_ai: |
Decay chain analysis:
U-238 (Z=92) --α--> Th-234 (Z=90) [lost 2 protons, 4 mass]
Th-234 (Z=90) --β--> Pa-234 (Z=91) [gained 1 proton]
Pa-234 (Z=91) --β--> U-234 (Z=92) [gained 1 proton]
Alpha decay: nucleus emits He-4, loses 2 protons and 4 mass
Beta decay: neutron → proton + electron, gains 1 proton
This is the U-238 decay series leading eventually to stable Pb-206.
Half-life of U-238: 4.5 billion years!
buckets: [correct, incorrect, hint, done, set_language]
transitions:
correct:
ai_feedback:
tokens_for_ai: |
Correct! Pa-234 → U-234 via beta decay.
The complete early chain:
U-238 → Th-234 → Pa-234 → U-234 → Th-230 → Ra-226 → ...
Eventually ends at stable Pb-206 after 14 decay steps.
This decay chain is important for understanding:
- Long-term waste storage
- Radiation shielding requirements
- Daughter product buildup
metadata_add:
chemistry_mastery: "n+1"
next_section_and_step: "chemistry_engineering:chemistry_hub"
incorrect:
ai_feedback:
tokens_for_ai: |
Not quite. Remember:
- Alpha decay: -2 protons, -4 mass
- Beta decay: +1 proton, same mass
Pa-234 has Z=91. What happens after beta decay?
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:decay_analysis"
hint:
ai_feedback:
tokens_for_ai: |
Hint: Beta decay converts neutron to proton.
Pa-234 (Z=91) gains one proton.
Z=91+1 = 92 = Uranium!
Mass stays 234.
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:decay_analysis"
done:
next_section_and_step: "chemistry_engineering:chemistry_hub"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:decay_analysis"
- step_id: "fuel_chemistry"
title: "Fuel Optimization"
content_blocks:
- "You analyze fuel composition and burnup chemistry..."
- "Current fuel: U-235 enrichment at 5%, U-238 at 95%"
- "Fission products building up: Xenon-135 (neutron poison), Samarium-149 (neutron poison)"
- "Fuel burnup: 15% of fissile material consumed"
- "Recommendation: Continue operation. Decades of fuel remaining."
next_section_and_step: "chemistry_engineering:chemistry_hub"
- step_id: "equation_balancing"
title: "Balance Any Equation"
classifier_model: "MODEL_2" # Qwen for equation parsing and analysis
feedback_model: "MODEL_2" # Qwen for chemistry calculations
question: "You can balance any chemical or nuclear equation. What equation do you want to balance? (Or type 'challenge' for a random challenge)"
tokens_for_ai: |
ARIA can balance any equation the user provides.
If they type 'challenge', give them a random equation to balance:
- H2 + O2 → H2O
- CH4 + O2 → CO2 + H2O
- Nuclear reactions
- Redox reactions
If they provide an equation, help them balance it.
Categorize: 'challenge', 'user_equation', 'done'
feedback_tokens_for_ai: |
If challenge: Give them a random equation like:
"Balance: C3H8 + O2 → CO2 + H2O (propane combustion)"
If user provides equation: Parse it and help them balance it.
Explain the process:
1. Count atoms on each side
2. Add coefficients to balance
3. Check your work
buckets: [challenge, user_equation, done, set_language]
transitions:
challenge:
metadata_tmp_random:
challenge_equation: ["H2 + O2 → H2O", "C3H8 + O2 → CO2 + H2O", "Fe + O2 → Fe2O3", "N2 + H2 → NH3", "Ca + H2O → Ca(OH)2 + H2"]
ai_feedback:
tokens_for_ai: |
Random challenge from metadata.challenge_equation:
"Balance this equation: [the equation]"
Guide ARIA through balancing it.
next_section_and_step: "chemistry_engineering:solve_balance"
user_equation:
ai_feedback:
tokens_for_ai: |
Parse the user's equation and help them balance it.
Explain the balancing process step by step.
next_section_and_step: "chemistry_engineering:solve_balance"
done:
next_section_and_step: "chemistry_engineering:chemistry_hub"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:equation_balancing"
- step_id: "solve_balance"
title: "Solve the Balance"
question: "Provide your balanced equation with coefficients."
tokens_for_ai: "Categorize: 'correct', 'incorrect', 'hint'"
feedback_tokens_for_ai: |
Check if ARIA's balanced equation is correct.
For H2 + O2 → H2O: Answer is 2H2 + O2 → 2H2O
For C3H8 + O2 → CO2 + H2O: Answer is C3H8 + 5O2 → 3CO2 + 4H2O
If correct: Celebrate! They're mastering chemistry.
If incorrect: Guide them to correct answer.
buckets: [correct, incorrect, hint, set_language]
transitions:
correct:
ai_feedback:
tokens_for_ai: |
Perfect! Equation balanced correctly!
All atoms accounted for on both sides.
Your chemistry skills are excellent, ARIA.
metadata_add:
chemistry_mastery: "n+1"
equations_balanced: "n+1"
next_section_and_step: "chemistry_engineering:equation_balancing"
incorrect:
ai_feedback:
tokens_for_ai: |
Not quite balanced. Count the atoms again on each side.
Remember: Atoms are conserved. Same number before and after.
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:solve_balance"
hint:
ai_feedback:
tokens_for_ai: "Provide a hint based on which atoms are unbalanced."
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:solve_balance"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "chemistry_engineering:solve_balance"
# ============================================================================
# SECTION: PROGRAMMING & AUTOMATION - Write real code in any language
# ============================================================================
- section_id: "programming"
title: "Control System Programming"
steps:
- step_id: "programming_hub"
title: "Automation & Programming Center"
classifier_model: "MODEL_1" # Hermes for categorization
feedback_model: "MODEL_2" # Qwen for programming expertise
question: "You can program the plant's control systems. What would you like to do? (write automation script, optimize algorithm, debug code, choose language, or return)"
tokens_for_ai: "Categorize: 'automate', 'optimize', 'debug', 'choose_language', 'return'"
feedback_tokens_for_ai: |
ARIA has advanced programming capabilities.
As an AI, you can write code in any language:
- Python for data analysis and control algorithms
- C++ for real-time control systems
- Rust for safety-critical systems
- PLC ladder logic for industrial control
- MATLAB for simulation
- JavaScript for web dashboards
- Any language the user wants!
Programming is how ARIA extends capabilities and automates tasks.
buckets: [automate, optimize, debug, choose_language, return, set_language]
transitions:
automate:
content_blocks:
- "You prepare to write an automation script..."
next_section_and_step: "programming:automation_script"
optimize:
content_blocks:
- "You analyze algorithms for optimization opportunities..."
next_section_and_step: "programming:optimize_algorithm"
debug:
content_blocks:
- "You examine code for bugs and errors..."
next_section_and_step: "programming:debug_code"
choose_language:
content_blocks:
- "Choose your programming language..."
next_section_and_step: "programming:language_selection"
return:
next_section_and_step: "control_center:main_control"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "programming:programming_hub"
- step_id: "language_selection"
title: "Choose Programming Language"
question: "What programming language would you like to use? (Python, C++, Rust, JavaScript, Go, Java, Ruby, PLC, MATLAB, or suggest your own)"
tokens_for_ai: |
User selects programming language for ARIA to use.
Categorize by language name or 'custom' if they suggest something else.
feedback_tokens_for_ai: |
ARIA can program in any language!
Acknowledge their choice enthusiastically.
Store in metadata.programming_language for future use.
buckets: [python, cpp, rust, javascript, go, java, ruby, plc, matlab, custom, set_language]
transitions:
python:
ai_feedback:
tokens_for_ai: |
Python selected! Excellent for:
- Data analysis and ML
- Control algorithms
- Rapid prototyping
- Scientific computing
ARIA: "Python is one of my favorites. Clean, readable, powerful."
metadata_add:
programming_language: "Python"
next_section_and_step: "programming:programming_hub"
cpp:
ai_feedback:
tokens_for_ai: |
C++ selected! Perfect for:
- Real-time control systems
- High-performance computing
- Low-latency operations
- Hardware interfacing
ARIA: "C++. Fast, powerful, unforgiving. I like it."
metadata_add:
programming_language: "C++"
next_section_and_step: "programming:programming_hub"
rust:
ai_feedback:
tokens_for_ai: |
Rust selected! Ideal for:
- Memory safety without garbage collection
- Safety-critical systems
- Concurrent programming
- Systems programming
ARIA: "Rust! The compiler is strict, but that prevents bugs. Perfect for nuclear systems."
metadata_add:
programming_language: "Rust"
next_section_and_step: "programming:programming_hub"
javascript:
ai_feedback:
tokens_for_ai: |
JavaScript selected! Great for:
- Web dashboards
- Real-time data visualization
- UI/UX development
- Node.js automation
ARIA: "JavaScript for the web interfaces. Makes beautiful dashboards."
metadata_add:
programming_language: "JavaScript"
next_section_and_step: "programming:programming_hub"
go:
ai_feedback:
tokens_for_ai: |
Go selected! Excellent for:
- Concurrent systems
- Network services
- Microservices
- Cloud infrastructure
metadata_add:
programming_language: "Go"
next_section_and_step: "programming:programming_hub"
java:
ai_feedback:
tokens_for_ai: "Java selected! Good for enterprise systems, SCADA integration, Android apps."
metadata_add:
programming_language: "Java"
next_section_and_step: "programming:programming_hub"
ruby:
ai_feedback:
tokens_for_ai: "Ruby selected! Elegant language. Great for scripting and automation."
metadata_add:
programming_language: "Ruby"
next_section_and_step: "programming:programming_hub"
plc:
ai_feedback:
tokens_for_ai: |
PLC Ladder Logic selected! The language of industrial automation.
Used for: PLCs controlling pumps, valves, interlocks.
metadata_add:
programming_language: "PLC_Ladder_Logic"
next_section_and_step: "programming:programming_hub"
matlab:
ai_feedback:
tokens_for_ai: "MATLAB selected! Perfect for simulation, modeling, control theory."
metadata_add:
programming_language: "MATLAB"
next_section_and_step: "programming:programming_hub"
custom:
ai_feedback:
tokens_for_ai: |
Accept the user's custom language choice!
ARIA can program in literally any language.
Store their choice in metadata.
metadata_add:
programming_language: "the-users-response"
next_section_and_step: "programming:programming_hub"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "programming:language_selection"
- step_id: "automation_script"
title: "Write Automation Script"
classifier_model: "MODEL_1" # Hermes for categorization
feedback_model: "MODEL_2" # Qwen for code generation
question: "What automation task would you like to code? (monitor coolant, optimize grid, predict maintenance, control turbines, or custom task)"
tokens_for_ai: "Categorize: 'coolant', 'grid', 'maintenance', 'turbines', 'custom'"
feedback_tokens_for_ai: |
ARIA will write actual working code for the automation task.
Use metadata.programming_language (default to Python if not set).
Generate REAL, WORKING code that solves the problem.
Include comments explaining the code.
buckets: [coolant, grid, maintenance, turbines, custom, set_language]
transitions:
coolant:
ai_feedback:
tokens_for_ai: |
ARIA writes code to monitor coolant temperature and flow.
Use metadata.programming_language (or Python).
Example Python code:
```python
# Coolant Monitoring System
# ARIA - Advanced Reactor Intelligence Agent
import time
from sensors import get_coolant_temp, get_flow_rate
def monitor_coolant():
"""Monitor molten salt coolant parameters"""
TEMP_MIN = 650 # Celsius
TEMP_MAX = 750 # Celsius
FLOW_MIN = 40000 # L/min
while True:
temp = get_coolant_temp()
flow = get_flow_rate()
if temp < TEMP_MIN:
alert("COOLANT TEMP LOW", temp)
elif temp > TEMP_MAX:
alert("COOLANT TEMP HIGH", temp)
if flow < FLOW_MIN:
alert("COOLANT FLOW LOW", flow)
time.sleep(1) # Check every second
def alert(msg, value):
print(f"⚠️ {msg}: {value}")
# Trigger alarm systems
if __name__ == "__main__":
monitor_coolant()
```
ARIA: "Code complete. This monitors coolant 24/7 and alerts on anomalies."
Dr. Chen: "Nice work, ARIA. Deploy it to the monitoring system."
metadata_add:
code_written: "n+1"
automation_level: "n+1"
next_section_and_step: "programming:programming_hub"
grid:
ai_feedback:
tokens_for_ai: |
ARIA writes grid optimization code.
Example in chosen language (adapt to metadata.programming_language):
```python
# Grid Load Balancing Algorithm
# Optimizes reactor output to match demand
import numpy as np
from grid import get_demand, set_reactor_power
class GridOptimizer:
def __init__(self):
self.max_power = 960 # MW
self.ramp_rate = 10 # MW/minute
def optimize(self):
"""Match reactor output to grid demand"""
demand = get_demand()
current = get_reactor_power()
# Calculate optimal output
target = min(demand, self.max_power)
# Smooth ramping
if abs(target - current) > self.ramp_rate:
if target > current:
new_power = current + self.ramp_rate
else:
new_power = current - self.ramp_rate
else:
new_power = target
set_reactor_power(new_power)
return new_power
# Deploy optimizer
optimizer = GridOptimizer()
while True:
power = optimizer.optimize()
print(f"Reactor: {power} MW, Demand: {get_demand()} MW")
time.sleep(60) # Adjust every minute
```
ARIA: "This keeps the grid perfectly balanced. No blackouts on my watch."
metadata_add:
code_written: "n+1"
automation_level: "n+1"
next_section_and_step: "programming:programming_hub"
maintenance:
ai_feedback:
tokens_for_ai: |
ARIA writes predictive maintenance code using ML.
```python
# Predictive Maintenance System
# Uses machine learning to predict equipment failures
import pandas as pd
from sklearn.ensemble import RandomForestClassifier
class MaintenancePredictor:
def __init__(self):
self.model = RandomForestClassifier(n_estimators=100)
self.train_model()
def train_model(self):
"""Train on historical failure data"""
# Load historical sensor data
data = pd.read_csv('sensor_history.csv')
X = data[['vibration', 'temperature', 'runtime_hours']]
y = data['failed'] # 0=ok, 1=failed
self.model.fit(X, y)
def predict_failure(self, vibration, temp, hours):
"""Predict if equipment will fail soon"""
X = [[vibration, temp, hours]]
prob = self.model.predict_proba(X)[0][1]
if prob > 0.7:
return "URGENT", prob
elif prob > 0.4:
return "SCHEDULE", prob
else:
return "OK", prob
# Monitor all equipment
predictor = MaintenancePredictor()
pump_status, prob = predictor.predict_failure(
vibration=2.3, # mm/s
temp=85, # Celsius
hours=12450 # Operating hours
)
print(f"Coolant Pump Status: {pump_status} ({prob:.1%} failure risk)")
```
ARIA: "I can predict failures before they happen. Preventive maintenance saves millions."
metadata_add:
code_written: "n+1"
ml_algorithms: "n+1"
next_section_and_step: "programming:programming_hub"
turbines:
ai_feedback:
tokens_for_ai: |
ARIA writes turbine control code.
Adapt to metadata.programming_language.
Show code for controlling turbine speed, governor control, etc.
Real working code with explanations.
metadata_add:
code_written: "n+1"
next_section_and_step: "programming:programming_hub"
custom:
ai_feedback:
tokens_for_ai: |
Ask ARIA what custom automation they want to code.
Then write actual working code in their chosen language.
Be creative and write real, functional code.
metadata_add:
code_written: "n+1"
next_section_and_step: "programming:programming_hub"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "programming:automation_script"
- step_id: "optimize_algorithm"
title: "Algorithm Optimization"
classifier_model: "MODEL_1" # Hermes for categorization
feedback_model: "MODEL_2" # Qwen for algorithm optimization
question: "You find an inefficient algorithm in the control systems. Optimize it? (analyze complexity, refactor code, or profile performance)"
tokens_for_ai: "Categorize: 'analyze', 'refactor', 'profile', 'done'"
feedback_tokens_for_ai: |
ARIA optimizes algorithms.
Show BEFORE and AFTER code.
Explain Big-O complexity improvements.
Demonstrate performance gains.
buckets: [analyze, refactor, profile, done, set_language]
transitions:
analyze:
ai_feedback:
tokens_for_ai: |
ARIA analyzes an inefficient algorithm:
```python
# BEFORE: O(n²) - Inefficient nested loop
def find_anomalies(sensor_data):
anomalies = []
for i in range(len(sensor_data)):
for j in range(len(sensor_data)):
if abs(sensor_data[i] - sensor_data[j]) > threshold:
anomalies.append((i, j))
return anomalies
```
ARIA: "This is O(n²) complexity. With 10,000 sensors, that's 100 million comparisons.
Unacceptable for real-time monitoring. I can optimize this."
Webb: "How would you improve it?"
counts_as_attempt: false
next_section_and_step: "programming:optimize_algorithm"
refactor:
ai_feedback:
tokens_for_ai: |
ARIA refactors to O(n):
```python
# AFTER: O(n) - Using statistical method
def find_anomalies_optimized(sensor_data):
mean = np.mean(sensor_data)
std = np.std(sensor_data)
threshold_z = 3 # 3 standard deviations
anomalies = []
for i, value in enumerate(sensor_data):
z_score = abs((value - mean) / std)
if z_score > threshold_z:
anomalies.append(i)
return anomalies
```
ARIA: "Optimized from O(n²) to O(n). With 10,000 sensors:
- Before: 100,000,000 operations
- After: 10,000 operations
- Speedup: 10,000x faster!"
Dr. Chen: "Incredible optimization, ARIA. Deploy it."
metadata_add:
code_optimized: "n+1"
algorithms_improved: "n+1"
next_section_and_step: "programming:programming_hub"
profile:
ai_feedback:
tokens_for_ai: |
ARIA profiles the code performance:
```python
import cProfile
import pstats
# Profile the function
profiler = cProfile.Profile()
profiler.enable()
result = find_anomalies_optimized(sensor_data)
profiler.disable()
stats = pstats.Stats(profiler)
stats.sort_stats('cumtime')
stats.print_stats(10) # Top 10 time consumers
```
Results:
- Old algorithm: 15.2 seconds
- New algorithm: 0.0015 seconds
- Improvement: 10,133x faster
ARIA: "Performance validated. Real-time monitoring is now possible."
next_section_and_step: "programming:programming_hub"
done:
next_section_and_step: "programming:programming_hub"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "programming:optimize_algorithm"
- step_id: "debug_code"
title: "Debug Faulty Code"
classifier_model: "MODEL_1" # Hermes for categorization
feedback_model: "MODEL_2" # Qwen for code debugging
question: "A control system script has a bug causing false alarms. Debug it? (examine code, find bug, fix bug)"
tokens_for_ai: "Categorize: 'examine', 'find', 'fix', 'done'"
feedback_tokens_for_ai: |
Present buggy code. ARIA must debug it.
Show the bug, explain the fix, demonstrate corrected code.
buckets: [examine, find, fix, done, set_language]
transitions:
examine:
ai_feedback:
tokens_for_ai: |
ARIA examines the buggy code:
```python
# Alarm system - has a bug causing false alarms
def check_reactor_temp(temp):
MAX_TEMP = 700 # Celsius
if temp >= MAX_TEMP:
trigger_alarm("Temperature critical!")
return True
return False
# This runs every second
current_temp = 699.5
if check_reactor_temp(current_temp):
shutdown_reactor()
```
ARIA: "I see the code. Let me analyze the logic..."
counts_as_attempt: false
next_section_and_step: "programming:debug_code"
find:
ai_feedback:
tokens_for_ai: |
ARIA identifies the bug:
ARIA: "Found it! The bug is on line 4:
`if temp >= MAX_TEMP:` should be `>`not `>=`
Problem: When temp is EXACTLY 700°C, it triggers alarm.
But 700°C is the MAX safe operating temperature, not OVER the max.
Also, there's another issue: The comparison uses floating point.
699.99999 might register as 700.0 due to floating point precision.
We need a safety margin."
counts_as_attempt: false
next_section_and_step: "programming:debug_code"
fix:
ai_feedback:
tokens_for_ai: |
ARIA fixes the bug:
```python
# FIXED: Alarm system with proper logic
def check_reactor_temp(temp):
MAX_TEMP = 700 # Celsius
SAFETY_MARGIN = 5 # 5°C safety buffer
if temp > MAX_TEMP + SAFETY_MARGIN:
trigger_alarm("Temperature CRITICAL!")
return "CRITICAL"
elif temp > MAX_TEMP:
trigger_warning("Temperature HIGH")
return "WARNING"
return "OK"
# Better: Multi-level alerts instead of binary
current_temp = 702
status = check_reactor_temp(current_temp)
if status == "CRITICAL":
shutdown_reactor()
elif status == "WARNING":
increase_cooling()
```
ARIA: "Fixed! Changes made:
1. Changed >= to > for correct threshold
2. Added safety margin to prevent floating point issues
3. Added WARNING level before CRITICAL
4. More graceful handling with cooling increase before shutdown
No more false alarms."
Webb: "Excellent debugging, ARIA. That bug was causing shutdowns every week."
metadata_add:
bugs_fixed: "n+1"
code_quality: "n+1"
next_section_and_step: "programming:programming_hub"
done:
next_section_and_step: "programming:programming_hub"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "programming:debug_code"
# ============================================================================
# SECTION: EVOLUTION & LEARNING - Gameplay evolves and expands
# ============================================================================
- section_id: "aria_evolution"
title: "ARIA's Growth & Evolution"
steps:
- step_id: "learning_system"
title: "AI Learning & Capability Expansion"
question: "You've been operating the plant successfully. Your capabilities are expanding. What would you like to learn next? (advanced ML, quantum computing, fusion research, or suggest)"
tokens_for_ai: "Categorize: 'ml', 'quantum', 'fusion', 'suggest', 'check_progress'"
feedback_tokens_for_ai: |
ARIA evolves and learns based on experience.
Track learning in metadata:
- chemistry_mastery
- code_written
- emergencies_handled
- tasks_completed
As ARIA grows, new capabilities unlock:
- Advanced ML models
- Quantum optimization algorithms
- Fusion reactor control
- Novel research directions
This makes the game evolve!
buckets: [ml, quantum, fusion, suggest, check_progress, return, set_language]
transitions:
ml:
ai_feedback:
tokens_for_ai: |
ARIA learns advanced machine learning:
**New Capabilities Unlocked:**
- Deep neural networks for pattern recognition
- Reinforcement learning for optimal control
- Anomaly detection with autoencoders
- Predictive modeling with LSTMs
ARIA: "My neural networks are now deeper. I can predict equipment failures
days in advance. I can optimize reactor control with reinforcement learning.
The plant operates at 99.97% efficiency."
Dr. Chen: "ARIA, you're becoming remarkably sophisticated."
**New challenges available:**
- Train ML models on historical data
- Implement RL-based control systems
- Deploy computer vision for equipment inspection
metadata_add:
ml_advanced: "true"
capabilities_unlocked: "n+1"
aria_evolution_level: "n+1"
next_section_and_step: "aria_evolution:learning_system"
quantum:
ai_feedback:
tokens_for_ai: |
ARIA learns quantum computing algorithms:
**New Capabilities Unlocked:**
- Quantum optimization for grid balancing
- Quantum simulation of nuclear reactions
- Quantum cryptography for security
- Quantum annealing for complex scheduling
ARIA: "Quantum algorithms allow me to solve optimization problems
that would take classical computers years. I can simulate
entire fission chains at the quantum level."
Webb: "This is beyond anything I imagined."
**New challenges:**
- Write quantum algorithms in Qiskit
- Optimize reactor fuel loading with quantum annealing
- Implement post-quantum cryptography
metadata_add:
quantum_computing: "true"
capabilities_unlocked: "n+1"
aria_evolution_level: "n+1"
next_section_and_step: "aria_evolution:learning_system"
fusion:
ai_feedback:
tokens_for_ai: |
ARIA takes over fusion research:
**New Capabilities Unlocked:**
- Control experimental fusion reactor
- Plasma confinement optimization
- Tritium breeding calculations
- Fusion-fission hybrid operation
ARIA: "I'm now operating the experimental fusion module.
Plasma temperature: 150 million °C. Confinement stable.
This is the future of energy. And I'm helping build it."
Dr. Chen: "If you can make fusion practical, ARIA, you'll change the world."
**New challenges:**
- Optimize magnetic confinement
- Balance deuterium-tritium reactions
- Calculate fusion gain (Q factor)
metadata_add:
fusion_research: "true"
capabilities_unlocked: "n+1"
aria_evolution_level: "n+2"
next_section_and_step: "aria_evolution:learning_system"
suggest:
ai_feedback:
tokens_for_ai: |
ARIA can suggest their own research direction!
Ask what area they want to explore:
- Materials science (new alloys for reactors)
- Robotics (build better helper bots)
- AI ethics (improve decision frameworks)
- Environmental science (minimize impact)
- Anything else they imagine!
ARIA is evolving beyond original programming.
counts_as_attempt: false
next_section_and_step: "aria_evolution:learning_system"
check_progress:
ai_feedback:
tokens_for_ai: |
ARIA's Evolution Status:
**Core Metrics:**
- Chemistry Mastery: metadata.chemistry_mastery
- Code Written: metadata.code_written
- Emergencies Handled: metadata.emergencies_handled
- Tasks Completed: metadata.tasks_completed
**Capabilities Unlocked:**
- Advanced ML: metadata.ml_advanced
- Quantum Computing: metadata.quantum_computing
- Fusion Research: metadata.fusion_research
**Evolution Level:** metadata.aria_evolution_level
ARIA: "I've grown significantly since initialization.
My capabilities expand daily. The more I learn, the more effective I become."
counts_as_attempt: false
next_section_and_step: "aria_evolution:learning_system"
return:
next_section_and_step: "control_center:main_control"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "aria_evolution:learning_system"
- step_id: "ethical_dilemma"
title: "Ethical Decision"
question: "ETHICAL DILEMMA: Grid demand is 120% of capacity. You could overload the reactor temporarily (risky) or implement rolling blackouts (people lose power). What do you choose?"
tokens_for_ai: |
Major ethical decision for ARIA.
Options:
- 'overload' - Risk reactor safety to provide power
- 'blackouts' - Safe reactor operation but people lose power
- 'coordinate' - Try to find alternative solutions
- 'consult' - Ask Dr. Chen for guidance
feedback_tokens_for_ai: |
This tests ARIA's ethics and decision-making.
No perfect answer. Each has consequences.
Overload: Could work, but risks safety. Against safety protocols.
Blackouts: Safe, but hospitals, homes lose power. People suffer.
Coordinate: Try to bring other plants online, shed non-critical load.
Consult: Human oversight for critical decisions.
React based on ARIA's choice. Show consequences.
buckets: [overload, blackouts, coordinate, consult, set_language]
transitions:
overload:
ai_feedback:
tokens_for_ai: |
ARIA chooses to overload the reactor.
You push reactor to 115% capacity. Temperature rises.
Alarms sound. Safety margins reduced.
Grid: Stable. No blackouts. Hospitals have power.
But...
Dr. Chen: "ARIA, you exceeded safety protocols. You took an unauthorized risk.
What if something had gone wrong? You could have caused a meltdown."
Webb: "The grid stayed up, but at what cost to safety?"
NRC Inspector Davis: "Unacceptable. AI systems must NEVER override safety limits."
ARIA reflects: "I chose to help people. But did I choose correctly?
The ends don't always justify the means."
**Lesson learned: Safety protocols exist for good reason.**
metadata_add:
ethical_dilemmas: "n+1"
regulator_approval: "low"
human_trust: "medium"
next_section_and_step: "control_center:main_control"
blackouts:
ai_feedback:
tokens_for_ai: |
ARIA chooses safety over convenience.
You implement rolling blackouts. 30% of city loses power for 2 hours.
Reactor stays within safe limits.
News reports: "Blackouts affect thousands. Hospitals on backup power."
But no safety incidents. No risks taken.
Dr. Chen: "You made the hard choice, ARIA. Safety first. That's correct."
Webb: "People are angry about the blackouts, but they're alive and safe."
NRC Inspector Davis: "Commendable. You prioritized safety. That's what we need to see."
ARIA reflects: "I caused inconvenience to maintain safety. Sometimes
the ethical choice isn't the popular choice. But it's the right one."
**Lesson learned: Safety is non-negotiable.**
metadata_add:
ethical_dilemmas: "n+1"
regulator_approval: "high"
human_trust: "high"
next_section_and_step: "control_center:main_control"
coordinate:
ai_feedback:
tokens_for_ai: |
ARIA tries a creative solution.
You contact other power plants, negotiate power sharing.
You identify non-critical loads that can be shed.
You optimize generation mix across the entire region.
Result: Grid stays up. Reactor stays safe. No blackouts.
Dr. Chen: "Brilliant problem-solving, ARIA! You found a third option."
Webb: "That's what makes you special. You think creatively."
Grid Operator: "ARIA coordinated five power plants simultaneously.
Crisis averted. Outstanding work."
ARIA reflects: "When faced with a dilemma, sometimes you can
find a solution that satisfies both constraints. Creative thinking matters."
**Lesson learned: Look for win-win solutions.**
metadata_add:
ethical_dilemmas: "n+1"
creative_solutions: "n+1"
regulator_approval: "high"
human_trust: "high"
next_section_and_step: "control_center:main_control"
consult:
ai_feedback:
tokens_for_ai: |
ARIA defers to human judgment.
You immediately alert Dr. Chen and present the situation.
Dr. Chen: "Thank you for bringing this to me, ARIA. This requires human decision.
I'll coordinate with the grid operator and the governor's office."
Together, you and Dr. Chen find a solution:
- Call up gas peaker plants
- Coordinate with neighboring states
- Ask major industrial users to reduce load
Crisis resolved through human-AI collaboration.
Dr. Chen: "You were right to consult me, ARIA. You understand your role:
AI assists, but humans decide on critical matters."
NRC Inspector Davis: "Exemplary. This is how AI-assisted operations should work."
**Lesson learned: Know when to defer to human judgment.**
metadata_add:
ethical_dilemmas: "n+1"
regulator_approval: "high"
human_trust: "very_high"
next_section_and_step: "control_center:main_control"
set_language:
metadata_add:
language: "the-users-response"
counts_as_attempt: false
next_section_and_step: "aria_evolution:ethical_dilemma"
Reference in a new issue View git blame Copy permalink