agent-reach

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md instructs runtime fetching/reading of outsider-authored web/social content (e.g., “Read any web page” via curl -s "https://r.jina.ai/URL" and social platform searches), which would ingest arbitrary third-party free text into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs the agent to "fetch the install guide" at runtime from https://raw.githubusercontent.com/Panniantong/agent-reach/main/docs/install.md so the agent will retrieve external instructions that could include commands to execute, creating a high-confidence runtime dependency.