verification

Pass

Audited by Gen Agent Trust Hub on Jul 29, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides instructions for a verification workflow within a specific software project (OpenChatCut). It defines procedures for ensuring that edits are correctly applied through static checks, agent runs, and manual reviews.
  • [COMMAND_EXECUTION]: The skill mentions using npx tsc --noEmit as part of its 'L0' verification level. This is a standard development practice for TypeScript type checking and does not involve arbitrary or malicious command execution.
  • [DATA_EXPOSURE]: References to the file path /media/uploads/ are documented as the location for project media assets. This is an application-specific directory and does not represent exposure of sensitive system configuration or user credential files.
  • [PROMPT_INJECTION]: The 'Real Agent transcript check' section instructs the agent to analyze previous chat records to verify successful execution. While this involves processing untrusted data (previous interactions), the instructions are specifically designed for error detection and quality assurance rather than executing instructions found within the data, presenting no significant risk.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 29, 2026, 03:01 PM
Security Audit — agent-trust-hub — verification