blendops-help

Pass

Audited by Gen Agent Trust Hub on Nov 17, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill contains no patterns associated with bypassing safety guidelines or overriding system instructions. It uses restrictive "Required laws" (e.g., no-arbitrary-python-interface.md) and strict routing logic to prevent unauthorized actions.
  • [DATA_EXFILTRATION]: No sensitive file paths, hardcoded credentials, or network exfiltration patterns were detected. The skill primarily references local documentation and other internal skill files using relative paths.
  • [REMOTE_CODE_EXECUTION]: There are no commands that download or execute external code. While it mentions third-party tools like mcp-for-blender in the documentation section, it explicitly states it does not execute Python or run Blender runtimes itself.
  • [OBFUSCATION]: The content is clear and uses standard Markdown/JSON formatting. No Base64, zero-width characters, or homoglyph attacks were found.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input (user_request) for routing purposes. However, the risk is mitigated by its 'L0' capability tier (no execution privileges), strict output templates, and a 'downgrade-only' truth state policy that prevents the agent from making unsupported claims based on user input.
  • [DYNAMIC_CONTEXT_INJECTION]: No dynamic execution patterns (such as the !command syntax) are present in the skill instructions.
  • [PRIVILEGE_ESCALATION]: No privilege escalation attempts (e.g., sudo, chmod) were identified. The skill is designed for triage and information routing only.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 17, 2026, 12:21 AM