lovstudio-repo2docs

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md Step 6’s incremental authoring loop reads each unit’s file contents from the user-selected source folder (GitHub clone/local dir/current dir) and converts it into MDX pages, so outsider-authored text submitted via the GitHub URL path is ingested at runtime.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill runs remote code at runtime (scripts/scaffold_docs.py calls npx to fetch+execute create-fumadocs-app@latest) and explicitly clones user-provided GitHub repos which the agent then reads into its authoring context (see the git clone flow and the install URL https://github.com/lovstudio/repo2docs-skill), so remote content/packages can directly control agent behavior.