markdown-to-skill

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill requires reading and writing the full Markdown content (copying original file bodies and frontmatter into generated SKILL.md), so any API keys, tokens, or passwords present in the source files would be reproduced verbatim in the agent's output, creating an exfiltration risk.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill explicitly scans and ingests Markdown files from a user-specified directory (Steps 2–5: detecting domain-like top-level directories, Glob **/*.md, and reading file contents to extract headings/frontmatter and produce SKILL.md), including examples that show website-scoped directories like docs.fastly.com, so untrusted/public website content could be imported and turned into executable skills that influence agent behavior.