product-image-processor

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly reads arbitrary Google Sheets (SKILL.md Step 2 using mcp__google__sheets_values_get and README "Process images from any Google Sheet") and then downloads image URLs from the open web with curl (SKILL.md Step 4), meaning untrusted, user-provided content directly drives what the agent fetches and processes and could therefore influence subsequent tool actions.