axiom-apple-docs

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

• Third-party content exposure detected (low risk: 0.10). The agent skill instructs the LLM to navigate arbitrary user/attacker-accessible websites (sosumi.ai and developer.apple.com) and read web page content via WebFetch, which is an active search/fetch risk.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill instructs fetching documentation from the third-party unauthenticated domain sosumi.ai via WebFetch during runtime, representing an unverifiable third-party operational dependency with weak/unknown provenance.