nano-banana-antigravity

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/generate_image.py, the required runtime workflow ingests outsider-authored free text via the --prompt CLI argument (main() → generate_image() → request_body["request"]["contents"][...]["parts"]), which is then sent to the Gemini image model.

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). The repository contains a hardcoded OAuth client secret in scripts/generate_image.py. ANTIGRAVITY_CLIENT_SECRET ("GOCSPX-...") is a high-entropy literal that can be used to exchange refresh tokens/access tokens and is therefore a real secret. ANTIGRAVITY_CLIENT_ID (the long .apps.googleusercontent.com value) is an OAuth client identifier (public) and not considered a secret; GOOGLE_TOKEN_URL and endpoint URLs are not secrets either.