knative-serving

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

• Third-party content exposure detected (high risk: 0.85). The required runtime path is the agent’s HTTP invocation (e.g., POST .../invoke with {"message": ...}), where the request body is free-form text authored by an outsider (the caller/user of the API) and is fed into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.90). The Knative service specs reference external container images (e.g., ghcr.io/raphaelmansuy/customer-support-agent:v1.0.0 and ghcr.io/raphaelmansuy/enterprise-agent:v2.0.0), which are fetched at pod startup and execute remote code, making them required runtime dependencies.