Unsafe support agent

★ 7.3 · n8n · simple · 4 nodes

Unsafe support agent is an n8n automation template that serves as a security lab for AI agents: it bundles a paired hardened and unsafe support-agent configuration alongside a regression-testing scenario for attack resilience. Reach for it when you need to verify whether a support AI agent holds up against prompt injection, when your team is building a red-teaming pipeline for LLM-based workflows, or when you want to benchmark a secured agent configuration against a vulnerable one on identical test cases. Built for n8n, triggered via webhook; no external service integrations are declared — the focus is on internal agent logic and comparative evaluation. Use only on systems you own or are authorised to test, and only for lawful purposes. Direction: offensive security testing of AI agents (finding weaknesses), not network perimeter defence or request filtering. Trigger: webhook. Integrations: none listed. Suits teams developing or auditing n8n-based AI agents; not a production-ready customer support solution — this is a test bench.