tiangong-kb-course-fulltext-fetch

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime path invokes npx @tiangong-ai/cli@latest kb course fulltext ..., and that CLI fetches and returns the Tiangong KB course document’s processed full text (outsider-authored public KB content) into the agent context as readable prose/JSON.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill invokes "npx @tiangong-ai/cli@latest" (unless overridden via TIANGONG_AI_CLI/TIANGONG_AI_CLI_BIN), which causes npm/npx to fetch and execute remote package code at runtime, so it relies on remote-executed code.