truefoundry-logs

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

• Third-party content exposure detected (high risk: 1.00). The skill's references (references/manifest-schema.md) explicitly permit providing remote OpenAPI spec URLs for MCP servers and external agent_card_url for hosted A2A agents which are fetched at runtime and converted into MCP tools / influence agent behavior, meaning untrusted third-party specs or agent cards could be ingested and materially change tool capabilities or actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly states that remote OpenAPI specs (e.g. https://api.weather.example.com/openapi.json) and hosted A2A agent cards (e.g. https://research-agent.example.com/.well-known/agent.json) are fetched at runtime and converted into MCP tools / agent behavior, meaning external content can directly control agent capabilities.