ad-copy-compliance-review

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly accepts public landing-page URLs (the materialUrl field described in SKILL.md and in references/form-schema.json / references/skill.json, and the x-required-any-of message requiring materialText/materialFile/materialUrl) and instructs callers to provide accessible public links, which implies the service will fetch/ingest those third-party pages to perform the compliance review—exposing the agent to untrusted external content that could carry indirect prompt-injection instructions.

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

• Potentially malicious external URL detected (high risk: 0.90). The provided runner makes runtime POST calls to the external AI Skills API (e.g. https://ai-skills.ai/api/execute) and relies on that service (with an API key) to execute the skill and return instructions/results that directly control agent behavior, so this is a required runtime dependency that influences prompts/execution.