agentauditai-sdk 1.0.0

On-chain EU AI Act compliance SDK for AI agents — immutable audit logs, KYA registration, incident reporting, post-market monitoring.

agentauditai-sdk

On-chain EU AI Act compliance for AI agents — immutable audit logs, Know Your Agent (KYA) registration, incident reporting, and post-market monitoring across 5 EVM networks.

Enforcement deadline: August 2, 2026.

---

Installation

pip install agentauditai-sdk

---

Quick Start

1. Register an AI agent (Art. 13, 26 — KYA standard)

from agentauditai import AgentAuditClient

client = AgentAuditClient(
    api_key="your-api-key",
    base_url="https://your-agentaudit-api",
    network="base",
)

registration = client.register_agent(
    agent_id="42",
    name="customer-support-agent",
    model="claude-sonnet-4-6",
    network="base",
)

print(f"Registered: tx={registration.tx_hash}")
print(f"Articles covered: {registration.articles}")

2. Audit an agent action (Art. 12, 19 — record-keeping)

from agentauditai import AgentAuditClient

client = AgentAuditClient(api_key="your-api-key", network="arbitrum")

result = client.audit_action(
    agent_id="42",
    action="LLM_DECISION",
    data={
        "prompt": "Summarise the customer complaint",
        "response": "Refund approved for order #8821",
        "model": "claude-sonnet-4-6",
    },
    risk_level="HIGH",
)

print(f"Logged on-chain: {result.tx_hash}")
print(f"Audit ID: {result.audit_id}")
print(f"Articles triggered: {result.articles}")

3. Get compliance report (Art. 72 — post-market monitoring)

from agentauditai import AgentAuditClient

client = AgentAuditClient(api_key="your-api-key", network="base")

# Risk score
risk = client.get_risk_score(agent_id="42")
print(f"Risk level: {risk.level}  score: {risk.score}")
print(f"Status: {risk.compliance_status}")

# Full compliance report
report = client.get_compliance_report(agent_id="42")
print(f"Agent: {report.agent_name}")
print(f"Total actions logged: {report.total_actions_logged}")
print(f"Applicable articles: {report.applicable_articles}")
print(f"Compliance status: {report.compliance_status}")
for obligation in report.obligations:
    status = "PASS" if obligation["met"] else "FAIL"
    print(f"  [{status}] {obligation['article']} — {obligation['obligation']}")

---

EU AI Act Coverage

Article	Obligation	Method
Art. 9	Risk management system	get_risk_score()
Art. 11	Technical documentation	register_agent()
Art. 12	Record-keeping & audit logs	audit_action()
Art. 13	Transparency to users	register_agent()
Art. 14	Human oversight	audit_action() with action tagging
Art. 19	Conformity assessment logging	audit_action()
Art. 26	Deployer obligations (KYA)	register_agent()
Art. 72	Post-market monitoring	get_compliance_report()
Art. 73	Serious incident reporting	audit_action(action="REPORT_INCIDENT")

---

Supported Networks

Network	Chain ID
Base Mainnet	8453
Arbitrum One	42161
Optimism Mainnet	10
Polygon Mainnet	137
Mantle Mainnet	5000

---

Configuration

Parameter	Environment Variable	Default
api_key	AGENTAUDIT_API_KEY	—
base_url	—	http://localhost:3000
network	—	base
timeout	—	30

---

Links

• Website: getagentaudit.xyz
• PyPI: pypi.org/project/agentauditai-sdk
• npm: npmjs.com/package/@agentauditai/sdk

---

AgentAudit AI — a RunLockAI product