swift-testing

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

• Third-party content exposure detected (high risk: 0.70). The skill's SKILL.md "Fetching More Docs" step 3 explicitly instructs the agent to retrieve documentation from the public sosumi.ai Markdown mirror (e.g., https://sosumi.ai/documentation/testing/test) when local files or other installed skills don't contain the page, so the agent will fetch and read content from an open third‑party website that could influence its behavior.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs the agent at runtime to fetch missing documentation from the sosumi mirror (e.g. https://sosumi.ai/documentation/testing/test), which would pull remote Markdown into the model’s context and thus can directly influence prompts/responses.