claude

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly covers installing and managing skills from external GitHub repos or URLs (references/skills-cli.md: "skills add — Install skills from a GitHub repo or URL") and SKILL.md lists "installing, updating, or managing skills" as a trigger, meaning the agent can fetch and incorporate untrusted third-party skill code/content that could influence its actions.