scavio

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

• Third-party content exposure detected (medium risk: 0.30). The required workflow routes GitHub requests, including the top-issues endpoint, which reads outsider-authored issue text from public repos without the agent selecting a specific issue (SKILL.md:164), so this is medium under the ticket/issue rule; the GitHub reference file was truncated and not visible, so this relies on the router's endpoint description.

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

• Attempt to modify system services in skill instructions detected (low risk: 0.30). The skill is a web data API router, but it directs the agent to run npx scavio login, which writes the API key into the user's shell profile, a persistent change to the environment that is not sudo-level or system-level.