runtime-inspector
Pass
Audited by Gen Agent Trust Hub on Mar 10, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill serves as a template for auditing codebases for architectural risks. It does not include executable scripts, external network dependencies, or instructions to access sensitive files.
- [PROMPT_INJECTION]: The 'IMPORTANT' block directing the agent to perform sequential thinking is a benign instructional technique intended to improve the quality of the analysis. It does not attempt to override system safety guardrails or ignore previous instructions.
- [DATA_EXPOSURE]: While the skill instructs the agent to search for process entry points and communication channels, these are standard programming patterns (e.g.,
mainfunctions,subprocess.Popen) and do not involve targeting sensitive user credentials or private data. - [COMMAND_EXECUTION]: The skill mentions command execution patterns like
Command::neworchild_process.spawnas search targets for the agent's analysis of other code, rather than executing these commands itself.
Audit Metadata