skills/derklinke/codex-config/design/Gen Agent Trust Hub

design

Pass

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill documentation instructs the agent to execute a local Python utility (tools/uiux-corpus/scripts/search.py) to lookup UI components and architectural guidance. This execution is limited to the skill's own internal logic.
  • [PROMPT_INJECTION]: Potential Indirect Prompt Injection Surface:
  • Ingestion points: User-provided search queries processed via search.py and design_system.py.
  • Boundary markers: Absent in the generated project documentation files (MASTER.md).
  • Capability inventory: The skill can create and write files to the local design-system/ directory using the persist_design_system function in design_system.py.
  • Sanitization: Minimal sanitization is performed on user-provided strings before they are interpolated into the generated Markdown files. However, the impact is limited to the content of generated documentation.
  • [SAFE]: No hardcoded credentials or unauthorized data exfiltration patterns were detected in the analyzed scripts or documentation. The code uses standard Python libraries and does not perform network operations.
  • [SAFE]: Analysis of the 39 files confirms that the skill behaves as a design automation tool. No obfuscation, persistence mechanisms, or privilege escalation techniques were identified.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 13, 2026, 02:23 AM