BitterPillEngineering
Pass
Audited by Gen Agent Trust Hub on May 2, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill uses bash commands for voice notifications via a local endpoint (http://localhost:31337/notify) and for maintaining a local execution log in the agent memory directory (~/.claude/PAI/MEMORY/SKILLS/execution.jsonl). These side-effects are documented and serve status-tracking purposes.- [PROMPT_INJECTION]: As an auditing tool, the skill ingests and analyzes other instruction sets. While this creates a surface for indirect prompt injection, the skill is designed to evaluate these instructions for removal rather than executing them, and this capability is essential to its auditing function.- [SAFE]: Access to local configuration files such as settings.json and CLAUDE.md is performed to identify which instructions are loaded in the agent environment. These read operations are consistent with the skill stated purpose of performing a full system audit.
Audit Metadata