Summarize

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

• Third-party content exposure detected (high risk: 0.90). The skill's summarize.sh script and SKILL.md explicitly accept and fetch arbitrary web/Youtube URLs (extract_content uses curl and yt-dlp to retrieve page/subtitle content) and then feeds that untrusted third‑party content into the summarization model, allowing instructions in those pages to influence summaries and action items.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's runtime script uses curl to fetch arbitrary user-supplied web pages (e.g., curl -sL "$INPUT" which could be "https://evolink.ai/blog/deepseek-v4-release-window-prep") and injects that fetched content directly into the USER message sent to the Evolink API, meaning remote URL content fetched at runtime controls the model prompt.