gemini-interactions-api

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs that, when MCP is not available, it should fetch the official docs at runtime (https://ai.google.dev/gemini-api/docs/interactions.md.txt and https://ai.google.dev/gemmini-api/docs/deep-research.md.txt), which would inject external documentation into the agent context and thus directly control agent instructions.