skill-installer

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

• Third-party content exposure detected (medium risk: 0.30). In list-skills.py, when listing, the workflow fetches and ingests free-text JSON directory listings from a GitHub repo path via the GitHub Contents API (https://api.github.com/repos/{repo}/contents/{path}?ref={ref}), and those directory entries can be authored by outsiders.

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

• Potentially malicious external URL detected (high risk: 1.00). The install and listing scripts perform runtime network fetches of GitHub-hosted skill content (e.g., https://codeload.github.com/{owner}/{repo}/zip/{ref} and https://api.github.com/repos/{repo}/contents/{path}?ref={ref}), downloading repo files (including SKILL.md and code) which can directly control agent prompts or deliver executable code used as a required dependency.

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

• Attempt to modify system services in skill instructions detected (high risk: 0.70). The skill instructs agents to "request escalation" when running networked install scripts and even allows downloading/overwriting preinstalled .system skills, which encourages obtaining elevated privileges or modifying system-level files and thus can compromise the host state.