skills-cli

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md workflow and references/cli.md explicitly allow adding and installing skills from public GitHub URLs and other git remotes (e.g., via bunx skills add <source> and https://github.com/.../tree/...), and those installed skills are treated as executable instructions that the agent will run, so untrusted third‑party content can influence agent actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill repeatedly instructs the runtime command bunx skills add to fetch remote skill repositories (notably https://github.com/sjunepark/custom-skills/tree/main/skills and the GitHub shorthand vercel-labs/skills), and those fetched "skills" are executable instructions that will be installed at runtime and can directly control agent prompts or execute code.