1) Core Orchestration
Decision and routing layer.
Interprets objectives, chooses direct vs delegated action, and coordinates tools/models/agents.
2) Core Memory
Continuity and recall layer.
Retrieves durable context and stores meaningful outcomes so the system stays grounded over time.
3) Reflection & Evaluation
Quality control layer.
Compares outputs to goals, identifies misses, and emits structured lesson candidates.
4) Core Learning (Novel)
Behavior adaptation layer.
Converts validated lessons into behavior updates and tracks impact with rollback-safe evolution.
5) Core Governance
Risk and policy layer.
Enforces approval boundaries, classifies risk, and keeps actions and adaptations policy-aligned.
The Learning Loop
Do → Remember → Evaluate → Adapt → Govern → Repeat
This is the mechanism that makes Carapace improve cycle by cycle instead of remaining static.
How the skills interact
- Orchestration executes toward an objective.
- Memory supplies context and records outcomes.
- Reflection scores quality and creates lessons.
- Learning applies vetted improvements.
- Governance validates and constrains change.
What you get
Guide: architecture, contracts, interaction map, implementation workflow.
Pack: downloadable zip with the 5 skills prepared for faster integration.
Payment path will support Simplex first, with additional methods added in future updates.
Buy Pack (Coming Online)
This option will let buyers purchase the Core Skills zip package directly and receive the 5-skill bundle ready for import.
Initial checkout target: Simplex. Additional payment methods will be added as the storefront expands.