manage-providers

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

• Third-party content exposure detected (high risk: 0.70). High: during the required discover-and-persist workflow, _list_providers.py scans bundle directories for *_provider.py modules, imports them, and calls get_provider_declarations()—those provider modules are outsider-authored code/content that can return arbitrary free-text fields (e.g., description) which are then persisted to marshal.json and later read/returned by list-providers/other flows into the agent’s context.