AI Admissibility is an external pre-execution admission boundary for AI-driven workflows, authority-bearing agents, MCP/tool access, CI/CD jobs, cloud identity flows, and other high-impact automation.
It is not a scanner, monitor, audit log, or self-authorizing chatbot guardrail.
No Admission = No Execution.
The core question is:
Should this actor, with this intent, in this current context, receive authority to act?
Official site: https://ai-admissibility.com/
The website and GitHub repositories are public showcase, documentation, proof, and demonstration surfaces.
They are not used as commercial checkout infrastructure, credential issuers, hosted customer runtime, or customer production execution surfaces.
- AI Admissibility Boundary — architecture / proof / evaluation material:
https://github.com/pinfloyd/ai-admissibility-boundary - AI Admissibility Action — current public GitHub Marketplace evaluation Action:
https://github.com/pinfloyd/ai-admissibility-action - cnp-action — compatibility workflow slug retained for existing references; not the canonical repository for new evaluation installs:
https://github.com/pinfloyd/cnp-action - Agent + Boundary Demo — bounded demonstration material, not production runtime:
https://github.com/pinfloyd/ai-admissibility-agent-boundary-demo - Hosted Authority reference package — historical engineering / reference material, not the canonical installed runtime:
https://github.com/pinfloyd/ai-admissibility-hosted-authority
The canonical installed boundary is represented publicly through the official site and controlled demonstration paths. Public GitHub Actions are evaluation surfaces; they do not provide a customer-specific production no-bypass guarantee or a generally open authority endpoint.
For collaboration, research, integration, or deployment discussions:
governance@ai-admissibility.com
The public website and GitHub repositories remain demonstration surfaces; any real-world commercial or production arrangement is handled separately from them.
Pre-run policy is necessary. External admission is the stronger boundary.
Platform-native controls improve the executor. External admission separates execution from authority.
If execution can proceed without an external allow decision, the system has policy, but not external admission authority.
Surrogate Boundary Test: Can execution proceed without an external allow decision?
Learn more: