improve-content

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

• Third-party content exposure detected (high risk: 1.00). The skill's required workflow explicitly ingests public third-party content—Step 1: "Fetch the URL and read the full rendered page" and Step 2: "Google the primary keyword. Read the top 5 results"—so untrusted web pages and SERP results are read and used to drive rewrite decisions, creating clear exposure to indirect prompt injection.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly fetches the user-supplied target page URL at runtime ("URL of the page to improve") and uses that fetched page content as the primary input to drive analysis and rewriting, so an attacker-controlled external page could directly influence prompts/instructions.