Skip to content

Security for AI agents

Know what your AI agents are doing. Prove it to auditors.

Your AI agents access customer data, modify production systems, and make autonomous decisions — with zero paper trail. The EU AI Act is in force. Regulators are asking. AgenticAudit is the answer.

  • Up and running in 5 minutes

    One docker compose up, log your first event, see it classified and mapped to compliance frameworks.

    Quickstart

  • Works with your agents

    Claude Code, LangChain, Codex, Cowork — or any agent via the Python SDK and REST API.

    Integrations

  • Compliance out of the box

    Every action automatically maps to GDPR, EU AI Act, and SOC 2 articles. Generate audit-ready PDF reports.

    Concepts

  • API Reference

    Every endpoint documented. Request/response schemas, error codes, examples.

    API Reference

How it works

Your agent does something. AgenticAudit captures it, classifies the risk, detects personal data, and maps it to the compliance articles that matter. Automatically.

from agentaudit import AgentAudit

audit = AgentAudit(api_key="aa_live_xxxxx")

event = audit.log(
    agent_id="booking-agent-v2",
    action="access_customer_record",
    data={"customer_email": "maria@example.com"},
    reasoning="Customer requested booking modification",
)

print(event.risk_level)   # "medium"
print(event.pii_detected) # True
print(event.frameworks)   # {"gdpr": ["art_30"], "ai_act": ["art_14"]}

Works with

Agent Support Method
Claude Code ✅ Full Deterministic hooks, enterprise-enforceable
Cowork ✅ Full Plugin with hooks, marketplace deployment
LangChain ✅ Full Callback handler
CrewAI Roadmap Event hook (planned)
Codex Partial Transcript parsing
Any agent ✅ Full REST API / Python SDK