safety-critical-patterns
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides purely instructional content for code quality and safety-critical system auditing. No indicators of prompt injection, data exfiltration, or malicious obfuscation were found.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to audit external source code, which represents an ingestion surface for untrusted data. However, the instructions focus on objective rule checks (NASA Power of 10) and use a structured output format for violations. The ingestion point is the target code being audited, while capabilities are limited to reporting findings and writing to a specific local directory (
.review/findings.json). - [COMMAND_EXECUTION]: Rule 10 provides examples of standard linting commands such as
ruff check --select=ALL,mypy --strict, andtsc --strict. These are presented as informative examples of best practices for developer tooling and do not involve arbitrary command execution or shell injection.
Audit Metadata