nemoclaw-contributor-onboard-messaging-channel

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs the agent to read upstream docs/source at runtime (e.g., https://docs.openclaw.ai/channels/telegram and https://github.com/openclaw/openclaw/tree/main/extensions/telegram) and use that content to drive follow-up questions, so fetched content can directly control agent prompts.