huggingface-spaces

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill includes examples that embed secrets directly on the command line (e.g., --secrets KEY=val / hf spaces secrets set <id> KEY=val) and instructs using tokens for auth, which can lead an agent to ask for and echo secret values verbatim in commands or code — an exfiltration risk.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill includes an explicit runtime pip-install command that fetches and executes remote code from git+https://github.com/facebookresearch/demucs inside app startup (subprocess.run(... "git+https://github.com/facebookresearch/demucs")), which installs and runs external repository code at runtime.