gemini-api-dev

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

• Third-party content exposure detected (high risk: 0.85). The skill’s REQUIRED workflow instructs fetching public documentation pages at runtime (e.g., llms.txt and specific ai.google.dev/...md.txt pages) via web fetch tools, and those fetched free-text pages are then ingested into the agent’s LLM context—an outsider-authored source.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill explicitly instructs runtime fetching of the API discovery spec (https://generativelanguage.googleapis.com/$discovery/rest?version=v1beta) and the docs index (https://ai.google.dev/gemini-api/docs/llms.txt) and specific .md.txt pages, which are external resources fetched during runtime that can directly influence agent behavior and instructions.