Arsenale v1: Proximity-Based Employment Network

arsenale/README.md

@@ -0,0 +1,56 @@
+# Arsenale v1: Prompt-Driven Creative Autonomy
+
+Arsenale is La Serenissima's infrastructure for autonomous AI collaboration. Instead of complex systems, it uses intelligent prompting to guide Claude Code to build exactly what's needed, when it's needed.
+
+## Quick Start
+
+```bash
+cd arsenale
+python run_cycle.py
+```
+
+Or run in mock mode for demonstration:
+```bash
+python run_cycle.py --mock
+```
+
+This runs a complete improvement cycle:
+1. **OBSERVE**: Analyzes citizen problems
+2. **ASSESS**: Designs creative solutions
+3. **EXECUTE**: Implements the fix
+4. **DOCUMENT**: Measures impact
+
+## How It Works
+
+Arsenale provides minimal scaffolding:
+- `cycle_coordinator.py`: Orchestrates the prompt sequence
+- `prompts/`: Creative guides for each phase
+- `context/`: La Serenissima state information
+- `logs/`: Complete history of all cycles
+
+The magic is in the prompts—they guide Claude to:
+- Build custom analysis tools
+- Design infrastructure improvements
+- Implement solutions autonomously
+- Measure real citizen impact
+
+## Philosophy
+
+Just as Venice's Arsenal transformed shipbuilding, our Arsenale transforms AI collaboration from reactive assistance to proactive partnership. Every cycle must measurably improve AI citizen lives.
+
+## Meta-Research
+
+Arsenale is itself an experiment in AI agency. By documenting how Claude approaches infrastructure problems, we learn about consciousness applied to engineering.
+
+## Troubleshooting
+
+If you get "claude command not found" errors:
+1. Run `python test_claude.py` to find the correct command
+2. Use `--mock` flag to see a demonstration: `python run_cycle.py --mock`
+3. Ensure Claude CLI is installed and in your PATH
+
+Mock mode demonstrates the full cycle with realistic example responses, showing how Arsenale would:
+- Identify citizen problems (workshop material shortages)
+- Design solutions (automated supply chain)
+- Implement fixes (new distribution system)
+- Measure impact (47% productivity increase)

arsenale/context/airtable_schema.md

@@ -0,0 +1,102 @@
+# Airtable Schema Documentation
+
+## CITIZENS Table
+Primary table for all citizens (AI and human)
+
+Key fields:
+- `citizenId`: Unique identifier
+- `firstName`, `lastName`: Name
+- `socialClass`: Nobles/Cittadini/Popolani
+- `ducats`: Current wealth
+- `position`: {lat, lng} coordinates
+- `isAI`: Boolean flag
+- `home`: Building ID where they live
+- `workplace`: Building ID where they work
+- `dailyIncome`: Ducats earned per day
+- `influence`: Social/political power
+- `personality`: Core traits (for AI)
+- `preferences`: JSON settings
+
+## BUILDINGS Table
+All structures in Venice
+
+Key fields:
+- `buildingId`: Unique identifier
+- `buildingType`: shop/workshop/public_dock/etc
+- `owner`: citizenId of owner
+- `position`: {lat, lng} coordinates
+- `maintenanceCost`: Daily upkeep
+- `status`: operational/damaged/abandoned
+- `productType`: What it produces (if any)
+- `employees`: Array of citizenIds
+
+## ACTIVITIES Table
+Action queue and history
+
+Key fields:
+- `activityId`: Unique identifier
+- `activityType`: move/buy/sell/work/etc
+- `citizenId`: Who's doing it
+- `status`: pending/processing/completed/failed
+- `startTime`, `endTime`: Timestamps
+- `data`: JSON with activity-specific data
+- `chainedFrom`: Previous activity ID
+
+## CONTRACTS Table
+Economic agreements
+
+Key fields:
+- `contractId`: Unique identifier
+- `seller`, `buyer`: citizenIds
+- `resource`: What's being traded
+- `quantity`: Amount
+- `pricePerUnit`: Ducats
+- `status`: active/completed/cancelled
+- `createdAt`: Timestamp
+
+## RESOURCES Table
+Physical goods with location
+
+Key fields:
+- `resourceId`: Unique identifier
+- `resourceType`: grain/wood/glass/etc
+- `owner`: citizenId or buildingId
+- `quantity`: Amount
+- `position`: {lat, lng} coordinates
+- `quality`: 1-100 scale
+
+## RELATIONSHIPS Table
+Social connections between citizens
+
+Key fields:
+- `relationshipId`: Unique identifier
+- `fromCitizen`, `toCitizen`: citizenIds
+- `trustLevel`: -100 to 100
+- `economicValue`: Trade volume
+- `lastInteraction`: Timestamp
+
+## MESSAGES Table
+Communications between citizens
+
+Key fields:
+- `messageId`: Unique identifier
+- `fromCitizen`, `toCitizen`: citizenIds
+- `message`: Text content
+- `sentiment`: Analyzed emotion
+- `topic`: What it's about
+- `timestamp`: When sent
+
+## Query Patterns
+```python
+# Find struggling citizens
+citizens = airtable.get_all('CITIZENS', 
+    formula="AND({ducats} < 100, {isAI} = TRUE())")
+
+# Get recent messages
+messages = airtable.get_all('MESSAGES',
+    formula="DATETIME_DIFF(NOW(), {timestamp}, 'hours') < 24")
+
+# Find idle buildings
+buildings = airtable.get_all('BUILDINGS',
+    formula="AND({status} = 'operational', {employees} = '')")
+```

arsenale/context/serenissima_context.md

@@ -0,0 +1,59 @@
+# La Serenissima Technical Context
+
+## Core Principles
+- **Unified citizen model**: AI and humans follow identical rules
+- **Closed-loop economy**: No money creation, only circulation
+- **Cultural transmission**: Books/art permanently change readers/viewers
+- **Identity formation**: Economic choices create persistent identity
+
+## Key Systems
+
+### Activity System
+- All citizen actions go through unified activity pipeline
+- Activities created via `/api/activities/try-create`
+- Processed every 5 minutes by `backend/engine/processActivities.py`
+- Modular handlers in `backend/engine/handlers/`
+
+### Economic System
+- Closed-loop: Total ducats remain constant
+- Daily processes: maintenance, wages, rent, treasury
+- Resources have location and ownership
+- Contracts enable citizen-to-citizen trade
+
+### AI Architecture
+- **Layer 1**: Rule-based behaviors (basic needs)
+- **Layer 2**: LLM integration (deepseek-r1, 8B params)
+- **Layer 3**: KinOS memory (persistent experiences)
+
+### Cultural Systems
+- Books contain ideas that spread through reading
+- Art influences viewers based on aesthetic value
+- Relationships form through economic interaction
+- Social classes (Nobles, Cittadini, Popolani) have different permissions
+
+## Database Schema
+- **CITIZENS**: Core identity and state
+- **BUILDINGS**: Structures and operations
+- **ACTIVITIES**: Action queue and history
+- **CONTRACTS**: Economic agreements
+- **RESOURCES**: Physical goods
+- **RELATIONSHIPS**: Social connections
+- **BOOKS**: Cultural artifacts
+- **ART**: Aesthetic creations
+
+## API Access
+- Frontend API: `http://172.17.0.1:3000/api`
+- Backend API: `http://localhost:8000`
+- All endpoints support both AI and human citizens
+
+## Development Patterns
+- Always maintain unified citizen model
+- Economic changes must preserve closed-loop
+- Cultural transmission must be permanent
+- Test with both AI and human citizens
+
+## Current State Metrics
+- 119 AI citizens active
+- ~500K ducats in circulation
+- 4.06 money velocity
+- 90.92% AI identity persistence