url-reader

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md requires routing “concrete URLs the user wants read” through a public extraction service (defuddle.md) at runtime, which fetches outsider-authored webpage body text and returns it as Markdown/YAML that the agent then uses in its LLM context—an indirect prompt-injection exposure path.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly invokes a runtime fetch with curl to "https://defuddle.md/" (SKILL.md) to retrieve webpage content that is then used as the agent's input/instructions, so this external service directly controls prompts at runtime.