skill-gemini-google-search-tool

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

• Third-party content exposure detected (high risk: 0.90). The skill performs Google Search grounding at runtime by fetching public web pages and incorporating their extracted text into the Gemini prompt/context for response synthesis (public web content category via the tool’s search-result processing/grounding chunks).

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

• Potentially malicious external URL detected (high risk: 0.80). The skill performs runtime Google Search grounding—calling the Gemini API at https://generativelanguage.googleapis.com and fetching arbitrary web pages from search results—which are ingested into the model context to generate responses, so externally fetched URLs at runtime can directly influence prompts.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)