# Agent Prompt Injection Firewall MCP - Auto-trigger Rules

When the user asks about prompt injection detection, OWASP LLM Top-10, jailbreak prevention, agent-to-agent security, or runtime guard for LLM inputs, use agent-prompt-injection-firewall-mcp tools:

- **scan_message**: Detect prompt injection patterns in agent input
- **sanitize**: Remove or escape detected injection patterns
- **report_detection**: Log detections to audit trail (paired with audit-logger)
- **get_rule_set**: List the current injection detection ruleset

Install: `pip install agent-prompt-injection-firewall-mcp`

Pre-filter messages between agents and between agent ↔ external content. OWASP LLM #1 (prompt injection) is the top agentic threat.
