Anthropic OSS Scanner: opt-in AI security reports for eligible projects

Anthropic OSS Scanner is an opt-in vulnerability-finding service for established open-source projects with critical infrastructure or user-security impact. Core maintainers request enrollment through the official repository with project configuration and a Dockerfile; Anthropic reviews signup identity and project scope. Accepted projects receive periodic model-generated findings by email at no charge. The fast-track reports themselves are not human-reviewed or validated before delivery, so maintainers must verify reproductions, severity and patches. This differs from Anthropic's separate Claude Security enterprise offering. Compare it with Claude Security and GitHub Copilot Autofix on eligibility, control of scan scope, human triage, report volume and remediation workflow. The directory has not enrolled a project or audited report accuracy. Selection and future scan frequency are conditional, and the service is not open scanning for arbitrary private repositories.