Skip to content

Document full-lifecycle AI agent safety stack positioning #56

@safal207

Description

@safal207

Core positioning

Use this as a high-level umbrella framing for ProofPath and its relationship to CML, LTP, and model reasoning layers:

Сквозной стек безопасности AI-агентов:
от намерения и рассуждения модели
до проверенного исполнения, блокировки и аудита.

English version:

A full-lifecycle safety stack for AI agents:
from intent and model reasoning
to verified execution, blocking, and audit.

Supporting frame

Claude / GPT = reasoning layer
CML = causal meaning and responsibility
LTP = continuity, replay, and trace
ProofPath = execution boundary, approval, blocking, and audit

Strong short lines

AI agents should not only reason.
They should act through verifiable safety boundaries.
Models propose.
ProofPath gates.
LTP records.
CML explains.
Audit proves.

Why this matters

This framing explains how the projects complement each other instead of competing:

  • model reasoning proposes actions;
  • CML preserves causal meaning and responsibility;
  • LTP preserves thread continuity and replay;
  • ProofPath decides whether high-risk actions can execute;
  • audit logs preserve evidence after the decision.

Deliverables

  • Add this framing to docs/internet-action-layer.md or a new positioning doc.
  • Add a concise version to the README reviewer summary.
  • Add an English version for grant applications.
  • Add a diagram showing model → CML/LTP → ProofPath → audit.

Related

Metadata

Metadata

Assignees

No one assigned

    Labels

    documentationImprovements or additions to documentationenhancementNew feature or request

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions