Files
timmy-config/allegro/goap/QUICKSTART.md
2026-03-31 20:02:01 +00:00

147 lines
3.3 KiB
Markdown

# GOAP Autonomy System - Quick Start
## Installation
```bash
cd /root/allegro/goap
./setup.sh
```
## Running the System
### Option 1: Systemd Service (Recommended)
```bash
# Start the service
systemctl start goap-autonomy
# Enable auto-start on boot
systemctl enable goap-autonomy
# Check status
systemctl status goap-autonomy
# View logs
journalctl -u goap-autonomy -f
```
### Option 2: Manual Execution
```bash
cd /root/allegro/goap
# Run one autonomy cycle
python3 self_goap.py --once
# Run continuous autonomy loop
python3 self_goap.py --run
# Run in background
nohup python3 self_goap.py --run > /var/log/goap.log 2>&1 &
```
## Managing Autonomy
```bash
# Check system status
python3 self_goap.py --status
# Enable autonomy
python3 self_goap.py --enable
# Disable autonomy
python3 self_goap.py --disable
# Force plan for specific goal
python3 self_goap.py --plan system_health
```
## Goals
The system manages 13 goals across 5 categories:
| Category | Goals | Priority |
|----------|-------|----------|
| HUNGER | Resource acquisition, Efficiency | HIGH |
| SAFETY | System health, Security, Data integrity | CRITICAL |
| LEARNING | Knowledge, Skills, Adaptation | MEDIUM |
| SOCIAL | User engagement, Relationships, Collaboration | MEDIUM |
| GROWTH | Self-improvement, Exploration | LOW |
## Actions
14 actions available across categories:
- **System**: check_system_health, cleanup_resources, restart_service, backup_data
- **Knowledge**: research_topic, index_knowledge, learn_from_interaction
- **Skills**: practice_skill, acquire_new_skill
- **Social**: send_message, proactive_check_in, share_knowledge
- **Growth**: self_analysis, experiment
## How It Works
1. **State Collection** - Every 60 seconds, collects system metrics
2. **Goal Evaluation** - Updates goal satisfaction based on current state
3. **Planning** - Every 5 minutes, creates plans for top-priority goals
4. **Execution** - Schedules and executes plans with monitoring
5. **Learning** - Tracks success rates and adapts
## Files
- `/root/allegro/goap/autonomy.log` - Operation log
- `/root/allegro/goap/goals_state.json` - Goal history
- `/root/allegro/goap/executor_state.json` - Execution stats
- `/root/allegro/goap/self_state.json` - System config
## Troubleshooting
### Service won't start
```bash
# Check for Python errors
python3 self_goap.py --once
# Check permissions
ls -la /root/allegro/goap/*.py
# Check logs
journalctl -u goap-autonomy -n 50
```
### No plans found
This is normal if all goals are satisfied or current state doesn't match action preconditions. The system will keep trying.
### High CPU usage
The system runs in adaptive mode and will reduce activity under high load.
## Integration
The GOAP system reads from:
- Timmy metrics database (`/root/allegro/timmy_metrics.db`)
- Heartbeat logs (`/root/allegro/heartbeat_logs/`)
- System metrics (CPU, memory, disk)
## Development
### Adding a Custom Goal
```python
from goap import Goal, GoalCategory, GoalPriority
class CustomGoal(Goal):
def evaluate(self, world_state):
return 0.5 # Satisfaction level
```
### Adding a Custom Action
```python
from goap import Action, ActionResult
class CustomAction(Action):
async def execute(self, context):
# Do something
return ActionResult.success_result("Done!")
```
## Support
See `README.md` for detailed documentation.