agent-cli-toolkit

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly instructs the agent to read GitHub PRs/Issues and timelines using gh and gh-llm (e.g., "gh pr view ...", "gh llm pr view ...") which pulls untrusted, user-generated content from public GitHub that the agent will interpret and could change its actions.

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

• Potentially malicious external URL detected (high risk: 0.90). I flag https://vite.plus because the skill explicitly tells users to run curl -fsSL https://vite.plus | bash (a runtime installation step that fetches and executes remote code) as the installer for the 'vp' tool the skill depends on.