corridorkey-green-screen

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

• Third-party content exposure detected (high risk: 0.80). This skill's runtime workflow explicitly instructs downloading and loading third‑party model weights and modules from public sites (e.g., the Hugging Face model URL "https://huggingface.co/nikopueringer/CorridorKey_v1.0/..." and "uv run hf download ..." for GVM/VideoMaMa) which are untrusted external content the agent loads and uses for inference.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's installation step runs a remote installer via curl -LsSf https://astral.sh/uv/install.sh | sh, which fetches and executes remote shell code and is a required prerequisite (the "uv" package manager) for the skill.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.70). The prompt includes troubleshooting steps that instruct running privileged system commands (e.g., "sudo systemctl restart docker") and guidance to install/reconfigure system-level toolkits, which push an agent/user to modify system services and use elevated privileges.