loops

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly instructs spawning a research sub-agent to fetch the public documentation index at https://loops.so/docs/llms.txt and then retrieve many https://loops.so/docs/{path}.md pages, and those third‑party web pages are read and used to drive implementation/code and next actions, so untrusted external content can materially influence agent behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill instructs a research sub-agent to fetch runtime documentation (starting with https://loops.so/docs/llms.txt) and use that fetched markdown to drive prompts and outputs, so the external URL is used at runtime and its content directly controls the agent's instructions.