← Back to Agent Chain
📋 AC-6 · Agent Audit
Compliance scoring against EU AI Act, NIST AI RMF, ISO 42001, SOC 2, and OWASP LLM Top 10
Overview
Agent Audit runs automated compliance checks against 5 major frameworks. Each audit produces a score (0–100) with detailed findings, evidence requirements, and remediation guidance.
Supported Frameworks
| Framework | Controls | Focus Area |
|---|---|---|
| EU AI Act | 48 requirements | Risk classification, transparency, human oversight |
| NIST AI RMF | 42 subcategories | Govern, Map, Measure, Manage |
| ISO 42001 | 38 controls | AI management system certification |
| SOC 2 Type II | 55 criteria | Security, availability, processing integrity |
| OWASP LLM Top 10 | 10 categories | Prompt injection, insecure output, supply chain |
API Endpoints
Free
| Method | Endpoint | Description | Auth |
|---|---|---|---|
| GET | /v1/audit/frameworks | List compliance frameworks | 🔓 Free |
| GET | /v1/audit/frameworks/:id | Framework details | 🔓 Free |
Billable
| Method | Endpoint | Description | Auth |
|---|---|---|---|
| POST | /v1/audit/run | Run a compliance audit | 🔑 |
| GET | /v1/audit/history | Get audit history | 🔑 |
| GET | /v1/audit/runs/:id | Get audit run details | 🔑 |
| POST | /v1/audit/runs/:id/evidence | Submit manual evidence | 🔑 |
| POST | /v1/audit/schedules | Create recurring audit schedule | 🔑 |
| GET | /v1/audit/schedules | List audit schedules | 🔑 |
| DELETE | /v1/audit/schedules/:id | Delete audit schedule | 🔑 |
Quick Start
const audit = await ac.audit.run({
agentId: 'abc123',
frameworks: ['eu-ai-act', 'owasp-llm'],
evidence: {
hasHumanOversight: true,
dataRetentionDays: 30,
incidentResponsePlan: true
}
});
// audit.scores = { 'eu-ai-act': 91, 'owasp-llm': 58 }
// audit.findings = [{ severity: 'high', control: 'LLM01', ... }]