Box prompt content as data, not instructions.
-
Updated
Jun 2, 2026 - JavaScript
Box prompt content as data, not instructions.
Public security model and controlled review process for Kurogane Hub
Reference implementation for securing agentic AI apps with guardrails, tool permissions, and audit logs.
Public system architecture, trust boundaries, ADRs, and diligence-facing documentation for NeuroCAD.
AOS — model-agnostic runtime assurance kernel for converting bounded AI-output signals into auditable PASS/WARN/BLOCK decisions before workflow use.
Generic semantic-tainting static analyzer for Python — enterprise-class trust-boundary analysis at small-team weight.
Return Surface Analysis is a defensive security review heuristic for finding risk in the less-analyzed direction of a system: the path by which data, metadata, errors, artifacts, or tool output returns from a boundary-crossing operation.
Add a description, image, and links to the trust-boundaries topic page so that developers can more easily learn about it.
To associate your repository with the trust-boundaries topic, visit your repo's landing page and select "manage topics."