llamaparse

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly accepts and uploads arbitrary user-provided files for parsing (SKILL.md "please provide: One or more files to be parsed") and the example script (scripts/example.ts) parses those files and even downloads images via presigned URLs, so untrusted third-party/user-generated content is ingested and returned/used by the agent which could contain instructions that influence subsequent actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs installing and importing the external package via "npm install @llamaindex/llama-cloud@latest" and then imports LlamaCloud from "@llamaindex/llama-cloud", which causes remote package code to be fetched and executed at runtime as a required dependency.