wb

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

• Third-party content exposure detected (high risk: 0.80). The skill's workflow (SKILL.md) requires using Membrane to run actions and proxy requests against a W&B connection (e.g., membrane action run and membrane request CONNECTION_ID /path/to/endpoint), which fetches user-generated content from the public W&B service that the agent is expected to read and act on, exposing it to indirect prompt injection.