tidy

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill dispatches full file contents to a reviewer and requires the reviewer to emit exact 1–3 line snippets as old_string/new_string in its JSON (verbatim matches), which forces the LLM to output raw file content and can leak API keys/secrets present in files.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The skill explicitly reads and applies textual edits to "arbitrary source files" (no project-root restriction) and performs git operations including writes and checkouts, so it can modify system or config files (e.g. systemd units, ssh configs) and thereby change machine state even though it doesn't request sudo—this makes it capable of compromising the host state.