cf-edit

Warn

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Both the PowerShell and Python scripts execute a secondary validation tool ('cf-validate') located at a relative path outside the skill's directory ('../../cf-validate/'). This behavior introduces a dependency on the presence and integrity of scripts in external locations.
  • [COMMAND_EXECUTION]: The Python script employs 'import("re")' for dynamic module loading. While the target is a standard library, this pattern is often used to bypass static code analysis.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by interpolating untrusted user input and JSON definitions into XML configuration files without exhaustive validation.
  • Ingestion points: Parameters '-Value' and '-DefinitionFile' in 'scripts/cf-edit.ps1' and 'scripts/cf-edit.py'.
  • Boundary markers: Not utilized.
  • Capability inventory: The scripts have file system write access and the ability to execute shell commands.
  • Sanitization: Inconsistent use of XML escaping; some values are written directly to element inner text without sanitization.
Audit Metadata
Risk Level
MEDIUM
Analyzed
May 14, 2026, 06:02 PM
Security Audit — agent-trust-hub — cf-edit