constraint-finder

Pass

Audited by Gen Agent Trust Hub on Jul 17, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No security vulnerabilities or malicious patterns were identified. The skill's functionality is limited to text-based analysis and user interaction through the 'Read' and 'AskUserQuestion' tools.
  • [PROMPT_INJECTION]: While the skill establishes a specific persona (Theory-of-Constraints coach) and provides detailed behavioral constraints (e.g., refusing to optimize non-constraints), these are standard instructional guidelines for a specialized agent and do not attempt to bypass system safety protocols or override fundamental instructions.
  • [DATA_EXFILTRATION]: The skill does not contain any network-enabled commands or tools. It focuses on analyzing provided process flows without accessing sensitive system paths or credentials.
  • [REMOTE_CODE_EXECUTION]: There are no patterns involving the download or execution of remote scripts, package installations, or dynamic code generation.
  • [INDIRECT_PROMPT_INJECTION]: The skill has a minimal attack surface for indirect injection as it lacks powerful capabilities such as file-writing or code execution. Its primary purpose is to process user-supplied descriptions of business processes and provide logical advice.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 17, 2026, 02:04 PM
Security Audit — agent-trust-hub — constraint-finder