software-ux-research

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

• Third-party content exposure detected (high risk: 0.70). The skill is a community-installed “software-ux-research” research/ops toolkit, and its described purpose (“Web research”) strongly suggests it fetches public web content at runtime (outsider-authored pages/articles) and feeds the extracted prose into the agent’s LLM context via its web-research workflow.

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

• Potentially malicious external URL detected (high risk: 0.90). The SKILL includes runtime install commands (e.g., npx skills add vasilyu1983/AI-Agents-public --skill software-ux-research) which fetch and execute code from the upstream GitHub repo (https://github.com/vasilyu1983/AI-Agents-public), so remote content is fetched at runtime and can control the agent.