zai-cli

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md and references/examples.md explicitly describe real-time web search, web page-to-markdown extraction, and GitHub repo exploration (e.g., "Convert Web Page to Markdown" and "real-time web search"), which require fetching and interpreting arbitrary public web pages and repos, exposing the agent to untrusted third-party content that could carry indirect prompt injections.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill instructs running "npx zai-cli", which at runtime fetches and executes the zai-cli package from the npm registry (see https://www.npmjs.com/package/zai-cli), so remote code would be executed as a required dependency.