unity-cli

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). Documentation exposes multiple legitimate but high-risk features that permit remote/automated execution of arbitrary C# and persistent control of a local Editor (eval/command/MCP/shell), plus tools that modify client configs and drive a persistent headless Editor — capabilities that, if abused, enable remote code execution, backdoor-like control, and data exfiltration.

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

• Third-party content exposure detected (medium risk: 0.30). In “Machine/agent mode — unity shell --protocol ndjson”, the required runtime path ingests caller-provided JSON request fields (including command as raw free text) over stdio and tokenizes/executes them.

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

• Potentially malicious external URL detected (high risk: 1.00). The SKILL.md includes install commands that fetch and immediately execute remote scripts at runtime (curl ... | bash and irm ... | iex) from https://public-cdn.cloud.unity3d.com/hub/prod/cli/install.sh and https://public-cdn.cloud.unity3d.com/hub/prod/cli/install.ps1, which constitute runtime remote code execution.