using-petropowers
Pass
Audited by Gen Agent Trust Hub on Jul 3, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill contains strong directives such as 'ABSOLUTELY MUST' and 'not negotiable' to enforce its workflow, and states that it 'overrides default system prompt behavior'. These patterns serve legitimate operational purposes and do not target safety filters.
- [PROMPT_INJECTION]: The skill creates an indirect prompt injection surface by instructing the agent to prioritize instructions found in repository files (e.g., CLAUDE.md). This is a standard customization feature and is presented as a mechanism for user control. Evidence chain: 1. Ingestion points: CLAUDE.md, GEMINI.md, AGENTS.md, user requests (SKILL.md); 2. Boundary markers: Absent; 3. Capability inventory: Shell access, file operations, web tools, subagent dispatch (SKILL.md, references/*.md); 4. Sanitization: Absent.
- [SAFE]: No hardcoded credentials, malicious network operations, or unauthorized privilege escalation patterns were detected across the skill files.
Audit Metadata