defuddle

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md specifies Defuddle CLI usage to parse “user-provided” webpage URLs at runtime (defuddle parse --md), meaning outsider-authored free text from arbitrary web pages can be ingested for summarization/analysis.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill invokes the Defuddle CLI at runtime to "defuddle parse --md", meaning arbitrary external webpages (the "" parameter) are fetched during execution and their content is ingested into the agent context, which can directly influence prompts/output.