Anthropic releases open-source framework for AI vulnerability discovery
Original: Anthropic's open-source framework for AI-powered vulnerability discovery
Why This Matters
First major open-source AI framework for automated code vulnerability discovery
Anthropic launched defending-code-reference-harness, an open-source framework for autonomous vulnerability discovery and remediation using Claude AI. The tool provides skills for threat modeling, scanning, triage, and patching based on partnerships with security teams.
Anthropic has released an open-source reference implementation called defending-code-reference-harness on GitHub for autonomous vulnerability discovery and remediation with Claude AI. The framework includes skills for threat modeling, scanning, triage, patching, and an autonomous scanning harness that can be customized. The project is based on learnings from partnerships with security teams at several organizations since launching Claude Mythos Preview. The repository includes documentation, a blog post detailing best practices, and a companion cookbook for SDK-only implementation. The framework follows a recon → find → triage → report → patch workflow loop. The project has received 922 stars and 74 forks on GitHub, though Anthropic notes the repository is not actively maintained and not accepting contributions.