claude-blog-brain

This is a public-content sanitization and manifest-generation utility with no evident data exfiltration or malware behavior. Its main security risk is destructive filesystem deletion: the caller-controlled outputRoot is recursively removed, and the check does not prevent it from being an ancestor of brainRoot or another valuable directory. It also aborts rather than excluding selected private entries. Severity is moderate and depends on the supplied output path; no credentials or local paths are transmitted.