nhi-audit 0.1.5

Open-source auditor for Non-Human Identities and AI Agent attack surfaces in cloud environments

AgentSentry 🛡️

Open-source auditor for Non-Human Identities and AI Agent attack surfaces across AWS, Azure, GCP, GitHub, Kubernetes, and your local machine.

"45 machine identities for every 1 human. Almost none of them are governed."

AgentSentry discovers every IAM role, API key, service account, SSH key, and AI agent in your environment — builds an attack graph of their access relationships — and scores the blast radius if any identity is compromised, using a novel AI-Amplification Factor that quantifies how autonomous AI agents multiply attack surface.

---

Quick Start

pip install nhi-audit
agentsentry interactive

No cloud credentials needed to try it:

agentsentry scan mock          # full multi-cloud demo
agentsentry scan local         # scan this machine now

---

Installation

pip install nhi-audit                   # core (local scanner included)
pip install nhi-audit[aws]              # + AWS
pip install nhi-audit[azure]            # + Azure
pip install nhi-audit[gcp]              # + GCP
pip install nhi-audit[github]           # + GitHub
pip install nhi-audit[k8s]             # + Kubernetes
pip install nhi-audit[all-clouds]       # everything

Windows PATH fix (run once):

python -m agentsentry --install-path

---

Provider Setup

Provider	Setup	Command
Local	Nothing	agentsentry scan local
AWS	aws configure	agentsentry scan aws
Azure	az login	agentsentry scan azure
GCP	gcloud auth application-default login	agentsentry scan gcp
GitHub	set GITHUB_TOKEN=ghp_...	agentsentry scan github
K8s	kubectl config use-context	agentsentry scan k8s
AI Agents	Nothing	agentsentry scan agents --path .

---

All Commands

agentsentry interactive                      # guided provider picker (recommended)
agentsentry scan mock                        # demo, no credentials
agentsentry scan local --path ./myproject    # scan specific directory
agentsentry scan aws --visualize             # + interactive HTML attack graph
agentsentry scan aws --enrich                # + CISA KEV threat intel
agentsentry scan all                         # auto-detect + scan everything ready
agentsentry providers                        # check what's configured
agentsentry blast "ml-pipeline-executor"     # blast radius analysis

---

Risk Scoring: P×R×E×A

Risk = Privilege × Reachability × Exposure × AI-Amplification

CRITICAL ≥ 100  |  HIGH ≥ 50  |  MEDIUM ≥ 20  |  LOW < 20

The AI-Amplification Factor is a novel research contribution — the first formal quantification of how autonomous AI agents multiply the blast radius of a compromised identity.

---

Standalone Executable

No Python needed. Download from GitHub Releases:

Platform	File
Windows	agentsentry-windows.exe
macOS	agentsentry-macos
Linux	agentsentry-linux

---

Repository Structure

agent-sentry/
├── agentsentry/        ← CLI tool (Python, open-source)
├── website/            ← Marketing site (Next.js, Vercel)
└── paper/              ← Research paper (IEEE LaTeX)

---

Links

• Website & Docs: agent-sentry-beta.vercel.app
• PyPI: pypi.org/project/nhi-audit
• Issues: GitHub Issues

License: MIT — free forever.