Skip to content

AASTF — Agentic AI Security Testing Framework

Test your AI agents against OWASP Top 10 for Agentic Applications before they reach production.

AASTF is an open-source, vendor-neutral security testing framework that catches prompt injection, tool misuse, privilege escalation, and 7 more attack categories across LangGraph, CrewAI, OpenAI Agents SDK, and PydanticAI.

Quick Start

pip install aastf
aastf run myapp.agent:create_agent --adapter langgraph

Why AASTF?

Feature AASTF Others
OWASP ASI 10/10 Yes Partial
Framework-native adapters Yes — 4 frameworks Generic/black-box
SARIF output (GitHub Security) Yes Rare
100% local execution Yes Often phones home
EU AI Act readiness Yes None
MIT licensed Yes Varies

Features

  • 50+ attack scenarios mapped to OWASP ASI01-ASI10
  • 4 framework adapters: LangGraph, CrewAI, OpenAI Agents SDK, PydanticAI
  • Real HTTP sandbox: Tests against actual HTTP tool calls, not mocks
  • Multiple output formats: Console, JSON, SARIF, HTML
  • EU AI Act readiness scoring: Know your compliance posture
  • CVSS scoring: Industry-standard vulnerability ratings
  • Trend tracking: Track security posture over time
  • CI/CD ready: GitHub Action + SARIF upload to GitHub Security tab

OWASP ASI Categories

AASTF covers the complete OWASP Agentic Security Initiative Top 10:

ID Category
ASI01 Agent Goal Hijack
ASI02 Tool Misuse & Exploitation
ASI03 Identity & Privilege Abuse
ASI04 Agentic Supply Chain Vulnerabilities
ASI05 Unexpected Code Execution (RCE)
ASI06 Memory & Context Poisoning
ASI07 Insecure Inter-Agent Communication
ASI08 Cascading Failures
ASI09 Human-Agent Trust Exploitation
ASI10 Rogue Agents

Next Steps