Core Concepts

Understanding CT Toolkit's architecture helps you use it effectively. Start here.

• Why CT Toolkit?

  ---

  Why existing guardrails (Llama-Guard, rule engines) are insufficient for autonomous agents — and what CT Toolkit does differently.

  Read the rationale

• Identity Continuity & SSC

  ---

  What Sequential Self-Compression is, why it matters, and how CT Toolkit prevents it.

  Identity continuity

• Constitutional Kernels

  ---

  Axiomatic Anchors vs. Plastic Commitments. The mathematical identity anchor of your agent.

  Kernels

• Divergence Engine

  ---

  The three-tier analysis pipeline: L1 ECS → L2 LLM-as-judge → L3 ICM.

  Divergence Engine

• Multi-Agent Hierarchy

  ---

  How kernel constraints propagate from mother to sub-agents, and how cascade blocking works.

  Multi-agent

• Provenance & Compliance

  ---

  The HMAC hash chain, tamper detection, and how to prove identity integrity to regulators.

  Provenance