Red Teaming

Test gateway apps, agents, models and MCP servers against real attack techniques, then fix what breaks with hardened prompts and new detections.

LLM Intelligence AssessmentOpen in console

Console › Assessments › Red Teaming

Get going

Assess a gateway appRun suites against the app's configured model through the gatewayVet an MCP serverStatic, dependency and live tool-surface scanRed team an agentAttack any HTTP, WebSocket, or voice agent endpointCompare modelsRun the same attacks on 2 to 8 models

Everything in Red Teaming

9 pages

Get started2

• Red Teaming overview
• Reading a report

Assessments4

• LLM Intelligence Assessment
• MCP Threat Detection
• Agentic red teaming
• Model red teaming

Operate3

• Runs, findings and schedules
• Attack library
• Turn findings into detections