langchain-python-quickstart

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

• Third-party content exposure detected (high risk: 0.75). The required workflow instructs fetching and implementing the LangChain quickstart from a public web URL, so the runtime HTTP-fetched page content can be ingested into the LLM context as readable free text (public web content category).

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

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly instructs fetching and implementing the live docs at https://docs.langchain.com/oss/python/langchain/quickstart during runtime, so remote content would directly control the agent's prompts/instructions.