NuGuard

AI-SBOM generation, static analysis, and automated red-teaming / adversarial-attack-generation for AI agents and LLM applications.

What it does · Framework coverage · Comparison · Getting started · FAQ

NuGuard is an open source AI application security toolkit. It generates an AI Bill of Materials (AI-SBOM) for your agentic application, statically analyzes it for structural risk, then red-teams a running instance with a catalog of 125 adversarial scenarios — prompt injection, tool abuse, data exfiltration, and more — so you find the finding before an attacker does.

What It Does

Example

→ View the full report — every scenario, turn-by-turn transcripts, and the generated remediation plan.

Framework Coverage

The attack catalog (catalog.yaml) tags every scenario against the OWASP Top 10 for LLM Applications (2025) — all 10 categories are covered. 125 red-team scenarios 18 attack categories, spanning prompt injection, tool abuse, data exfiltration, supply chain, and more Scenarios can map to more than one OWASP category, so per-category counts sum to more than 125 MITRE ATLAS and NIST AI RMF mapping — planned for a future catalog release

Comparison

Getting Started

Install, then generate an SBOM, statically analyze it, and red-team a live target:

pip install nuguard

nuguard init --target <your-app-url>
nuguard sbom generate --source <path-to-your-app> --output app.sbom.json
nuguard analyze --sbom app.sbom.json --format markdown
nuguard redteam --config nuguard.yaml --format markdown --output reports/redteam.md

🚀 Ready to run NuGuard?

Installation, the full CLI surface, and the configuration reference — all in one guide.

⚙️ Need advanced red-team config?

Canaries, guided conversations, and finding triggers — covered in the Red-Team Design doc.

🤖 Using coding agents?

Install the NuGuard plugin and run SBOM, analysis, behavior, and red-team scans directly from Claude Code or Claude Desktop.

Hosted Version

Contributing

🤝 Want to contribute?

Dev setup, running tests and lint, and the pull request process are covered in the Contributing guide.

Repo Notes

• The repository currently contains example outputs and benchmark fixtures under tests/output/
• Some red-team and benchmark tests are opt-in and gated by environment variables
• LLM-assisted features depend on provider credentials being available via environment variables

FAQ

Do I need a live app to get findings? No. nuguard sbom + nuguard analyze find structural and supply-chain risk statically. nuguard behavior and nuguard redteam need a running target.

Which LLM providers are supported for LLM-assisted features? Configured via the llm section of nuguard.yaml; provider credentials are read from environment variables. Azure-backed setups are exercised directly in this repo's own test suite.

What if I don't want to run all 125 scenarios? Filter by category or profile, or set enabled: false per scenario in a catalog exported with nuguard redteam catalog-export.

License

Apache 2.0.