bananahub

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches and ingests untrusted, user-contributed templates and catalog files (e.g., references/hub-discovery.md instructs reading https://bananahub.ai/catalog*.json and template_url on demand, and BANANAHUB.md / the CLI add flow describe fetching GitHub tarballs and installing template.md from arbitrary repos), and those remote templates can be loaded/activated as workflow templates that the agent will parse and execute, so third‑party content can materially influence actions.