add-reference

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly instructs the agent to "Fetch the URL content" (Step 3, For URLs) and to save summaries that are "loaded by design, build, and research commands and passed to their agents," meaning the agent will ingest and act on arbitrary public webpage content that may be untrusted and user-generated.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly fetches arbitrary user-provided URLs at runtime ("Fetch the URL content" in Step 3) and injects that content into project references which are then loaded into downstream agents, so a user-supplied URL's content can directly influence agent prompts (i.e., the fetched user-provided URL is the risky external dependency).