agent-skill-update-monitor

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

• Third-party content exposure detected (high risk: 0.75). The required workflow includes fetching repository state and downloading/cloning from a (potentially user-supplied) GitHub repository at runtime (outsider-authored public web content), which can be ingested as readable text for the LLM to summarize changes.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill uses the repository URL https://github.com/lensetek/Book-Author-Agent-Skills at runtime to fetch/clone and install agent skills, and those fetched files (SKILL.md and agents) can directly control prompts or include executable scripts, so this external content is a runtime dependency that can change agent behavior.