langchain-typescript-quickstart

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

• Third-party content exposure detected (medium risk: 0.65). SKILL.md instructs fetching external LangChain live documentation over HTTP/Docs MCP (“Fetch that page … implement what it shows”), which is public outsider-authored free text that can be ingested into the LLM context at runtime when the agent reads the downloaded page contents.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly instructs the agent at runtime to "Fetch that page" and implement its contents, meaning the external URL https://docs.langchain.com/oss/javascript/langchain/quickstart would be fetched and used to control the agent's code/instructions.