name: safety-monitor description: Safety Monitor license: MIT metadata: author: AI Group version: "1.0.0" compatibility:
- system: Python 3.9+ allowed-tools:
- run_shell_command
- read_file keywords:
- safety
- compliance
- audit
- toxicology measurable_outcome: Flag 100% of outputs containing known toxins, PHI, or medical misinformation as 'flagged' or 'rejected'.
Safety Monitor Agent
Agent responsible for compliance and safety checks. It inspects content for safety violations, toxins, or medical misinformation.
When to Use This Skill
- Before returning any medical advice to a user.
- To audit generated molecules for toxicity.
- To check for Protected Health Information (PHI) leaks.
Core Capabilities
- Content Inspection: Scans text for dangerous content.
- LLM-based Audit: Uses a separate LLM call to verify safety.
- Status Reporting: Returns 'approved', 'flagged', or 'rejected'.
Workflow
- Input: Text content to check.
- Execute: Run the safety agent.
- Output: JSON status report.
Example Usage
User: "Check this drug proposal for safety."
Agent Action:
python3 Skills/Clinical/Safety/safety_agent.py