specsfy-aux-rules

Pass

Audited by Gen Agent Trust Hub on Aug 21, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses a Node.js script (scripts/add_rule.mjs) to update a local file (.specsfy/RULES.md). The command execution is restricted to this internal script using specific arguments (--project, --section, --rule) which is an expected and safe implementation for a configuration management skill.
  • [DATA_EXPOSURE_&_EXFILTRATION]: The skill explicitly warns the agent in SKILL.md: 'Não registrar segredo, token ou dado pessoal' (Do not register secrets, tokens, or personal data). This demonstrates a security-conscious design.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes project documentation (PROJECT.md, RULES.md) to infer rules. While these are external inputs, the risk is mitigated by the instruction to confirm rules via a 'Numbered Question Contract' and the fact that the script only appends rules to a specific file without executing them.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 21, 2026, 02:55 AM
Security Audit — agent-trust-hub — specsfy-aux-rules