skills/sfkislev/flue/Autodesk/Gen Agent Trust Hub

Autodesk

Pass

Audited by Gen Agent Trust Hub on May 30, 2026

Risk Level: SAFECREDENTIALS_UNSAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [CREDENTIALS_UNSAFE]: The file SKILL.md contains a hardcoded local path 'C:\Users\fredd.claude\skills\flue\SKILL.md', which reveals a specific OS username.
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions to install the 'flue' package from PyPI via 'pip install flue' as a prerequisite for its functionality.
  • [COMMAND_EXECUTION]: The skill invokes local Python scripts (e.g., 'adapters/3dsmax_adapter/3dsmax_bridge.py') through the shell to communicate with desktop applications.
  • [REMOTE_CODE_EXECUTION]: The architecture is designed to execute arbitrary script content, such as MAXScript, Python, or ExtendScript, within the runtime environments of professional desktop software.
  • [PROMPT_INJECTION]: The skill facilitates indirect prompt injection by interpolating user-provided instructions into scripts executed by host applications without explicit sanitization.
  • Ingestion points: User-directed automation requests in SKILL.md.
  • Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the script-generation flow.
  • Capability inventory: Shell command execution and script bridge execution within 3ds Max, Photoshop, and other apps (FLUE.md).
  • Sanitization: No mention of input validation or content escaping for the generated scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
May 30, 2026, 08:22 PM
Security Audit — agent-trust-hub — Autodesk