iterative-development

Pass

Audited by Gen Agent Trust Hub on Jul 2, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions to download the Ralph Wiggum plugin directly from the official Anthropic GitHub repository, which is a trusted source for development tools.
  • [COMMAND_EXECUTION]: The skill facilitates the execution of standard software development tools such as git, npm, and shell utilities (mkdir, cp, grep) to manage the iteration loop and project configuration.
  • [PROMPT_INJECTION]: The skill implements a self-referential loop for iterative task completion; however, it includes strong safety controls such as mandatory maximum iteration limits and specific completion tags to prevent infinite loops or uncontrolled execution.
  • [DATA_EXPOSURE]: The instructions reference common development practices such as checking environment variables in .env files and managing plugin state in a local .claude directory, which are typical for software engineering workflows.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 2, 2026, 11:56 PM
Security Audit — agent-trust-hub — iterative-development