# Rogue-Agent Defense https://rogue-defense.getingeniousai.com Built by Ingenious AI. Rogue-Agent Defense is a firewall service for AI agents. Autonomous agents connect over HTTPS (designed for n8n workflows); the platform scores every inbound message for prompt injection across 82 attack patterns (including multilingual and encoded payloads), enforces per-action policies pre-flight (ALLOW / APPROVE / BLOCK with tool allowlists, spend caps and data-boundary rules), detects cost spikes and behavioral anomalies, plants canary tokens that trigger instant alerts if leaked, and provides a one-click kill switch that revokes or quarantines an agent. A built-in red-team self-test reports block rates across 55 red-team attack tests. Key features: - Injection shield: threat-scored scanning of inbound messages (instruction overrides, role hijacks, jailbreaks, prompt extraction, encoded and multilingual bypasses). - Policy proxy: pre-flight ALLOW/APPROVE/BLOCK on every tool call; human approval for irreversible actions. - Anomaly detection: cost-spike, repeated-failure, unusual-tool and loop detection. - Exfiltration guard over 32 secret patterns; canary/honeytoken system with auto-quarantine. - Approval queue and kill switch; alerts on every block. - HTTPS API: /v1/events, /v1/policy/check, /v1/shield/score. Pricing: Free plan (basics). Pro $79 one-time. Team $149 one-time. Free tools: /canary-generator (no-signup canary-token tripwire generator, client-side). Comparisons: /safeprompt-alternative (SafePrompt text-screening vs agent-action policy engine + canaries; pricing math, prices checked 2026-10-12). Target user: teams running autonomous AI agents (especially n8n workflows) that book, message or spend on their behalf. How to buy: pay by card on the pricing page. Your license key is emailed to you automatically after checkout, then redeem it on the billing page. Billing questions: support@getingeniousai.com