devtool
Anthropic's open-source framework for AI-powered vulnerability discovery
This is a reference framework from Anthropic for evaluating AI models' ability to discover code vulnerabilities, offering specific value to AI security researchers.
6.3Overall
Utility6
Onboarding5
Craft6
Niche fit7
Longevity7
Five dimensions scored independently; overall is a weighted average. Scores are only comparable within this same rubric.
Good for
AI security researchers, machine learning engineers, especially those needing to evaluate or benchmark AI models for code vulnerability detection.
Not for
Developers looking for out-of-the-box code scanning tools, or security engineers not involved in AI model evaluation.
Project description
github.com
Alternatives
CodeQL (for static analysis, not AI evaluation)Semgrep (for static analysis, not AI evaluation)