livestorm

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

• Third-party content exposure detected (high risk: 0.80). The skill's SKILL.md explicitly instructs the agent to run Membrane actions and to proxy arbitrary requests to the Livestorm API (see "Running actions" and "Proxy requests" / membrane request), which ingests user-generated third‑party data (participants, recordings, event descriptions) that the agent is expected to read and could materially influence subsequent actions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill requires installing and running the Membrane CLI via "npm install -g @membranehq/cli", which fetches remote code from the npm registry and is executed/relied upon at runtime, so it can directly introduce and run external code.