market-research-digest

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

• Third-party content exposure detected (high risk: 0.90). The skill's Prompt Template and Setup explicitly instruct the agent to "search the web" and use a Browser for "web signals" and to read links in Google Drive/Granola files (SKILL.md Prompt Template and Setup), which ingests untrusted public web and user-generated content that can influence summaries and next actions.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill explicitly reads files from Google Drive (drive.google.com), queries Granola meeting notes, and follows arbitrary web links at runtime to inject those external contents into the model context to synthesize the digest, so these fetched resources can directly control prompts and are required for the skill.