vercel-logs

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

• Third-party content exposure detected (high risk: 0.75). The required workflow runs the Vercel CLI to fetch production log output (outsider-authored data from external systems/users), which is then filtered (e.g., grep -E '^[{]’ | jq) and thus becomes readable free text/JSON in the agent context at runtime; additionally, log lines can contain arbitrary message content that could include prompt-injection strings.