ant-design

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

• Third-party content exposure detected (high risk: 0.75). SKILL.md:10-12 instructs refreshing the live catalog via curl from a public outsider-controlled URL, whose fetched design.md text would be read/ingested by the agent at runtime, creating an indirect prompt-injection surface.

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

• Potentially malicious external URL detected (high risk: 0.90). The SKILL.md explicitly instructs a runtime fetch using curl to https://api.agent-ds.oday-bakkour.com/v1/systems/ant-design/design.md, which would supply the DESIGN/Agent Prompt Guide and token data that can directly control agent prompts.