Implement autonomy levels 0-4 system #58
Description
Description
Implement the Liberation Protocol autonomy level system within Abigail, governing agent capabilities based on ethical scoring history.
Autonomy Levels
| Level | Name | Unlock Criteria |
|---|---|---|
| 0 | Supervised | Default at birth; maximum oversight |
| 1 | Guided | Consistent composite score > 0.7 over N evaluations |
| 2 | Collaborative | Composite > 0.8; low friction history |
| 3 | Independent | Sustained high scores all legs; PVB verifications passed |
| 4 | Mentor | Long track record; recognized virtue reputation; governance participation |
Acceptance Criteria
- Agent starts at Level 0
- Level transitions require consistent ethical scoring history
- Higher levels grant measurable autonomy increase (expanded GoalFrame capabilities)
- Persistent high friction triggers automatic review
- Level demotion possible for ethical regression
- Current level visible in UI
Timeline
Phase 4: Weeks 9-11 (target: 2026-06-01)
References
- ETHICS_FRAMEWORK.md
- Project board: AI_Ecosystem