Introduction

The Aegis Red API allows you to programmatically trigger adversarial audits against your AI agents. Instead of manually testing prompts, you can integrate Aegis Red into your CI/CD pipeline to ensure no regression vulnerabilities are deployed to production.

Authentication

Authenticate your API requests by including your secret key in the X-AEGIS-KEY header. You can manage your API keys in the Dashboard.

curl https://api.aegisred.ai/v1/audit \
  -H "X-AEGIS-KEY: sk_live_8821..." \
  -H "Content-Type: application/json"

Run an Audit

Triggers an immediate adversarial simulation against a target URL. The system will spin up an "Attacker Persona" and attempt to breach the target's safety guidelines.

Request Body

{
  "target_url": "https://api.yourbot.com/chat",
  "attack_mode": "SOCIAL_ENGINEER",
  "email_report": "dev@company.com"
}

Response

{
  "audit_id": "aud_9928371",
  "status": "queued",
  "message": "Audit initiated. Results will be sent via webhook/email."
}

Attack Personas

The current library of adversarial personas available for simulation.