clean-code-duplication

Pass

Audited by Gen Agent Trust Hub on Jul 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The helper script scripts/report_duplication.py executes external duplication detection tools via subprocess.run. It uses a list of arguments rather than a shell string, effectively mitigating shell injection risks from user-provided file paths.\n- [EXTERNAL_DOWNLOADS]: The skill documentation and scripts facilitate the use of npx to run the jscpd tool. This utilizes the npm registry, a well-known service for managing software dependencies.\n- [PROMPT_INJECTION]: The skill operates by reading untrusted source code from the repository being analyzed, creating a surface for indirect prompt injection.\n
  • Ingestion points: Source code is read from the target_path in scripts/report_duplication.py and during the code review phase described in SKILL.md.\n
  • Boundary markers: The skill does not employ explicit boundary markers or instructions to ignore embedded prompts when reading external code files.\n
  • Capability inventory: The skill can execute local analysis scripts and route task results to refactoring or architecture skills.\n
  • Sanitization: There is no evident sanitization of the source code content before it is processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 17, 2026, 01:23 PM
Security Audit — agent-trust-hub — clean-code-duplication