Kakunin

Kakunin is KYC compliance infrastructure for AI agents. It issues X.509 identities, monitors agent behavior in real time, and generates audit trails designed for regulatory review. You can delegate scoped authority, score what an agent does and says, and prove control on recognized compliance frameworks.

Designed for regulated workflows, Kakunin helps map agent activity to specific regulatory clauses and produces compliance reports built from append-only, WORM-style audit logging. With features like risk scoring, alerts, and auto-revocation behavior, you can respond when an agent crosses defined thresholds.

Kakunin is API-first and supports REST and a TypeScript SDK, with an OpenAPI 3.0 spec and HMAC-signed webhooks. Sandbox mode, webhook verification helpers, and retry/queue handling are included to help you certify agents and stream events safely into your compliance processes.

What does Kakunin provide for AI agents?

Kakunin provides cryptographic identities (X.509), real-time behavioral monitoring, and regulator-ready audit trails, plus scoped authority to control what an agent is allowed to do.

How does agent certification work in Kakunin?

You create an agent with permitted actions and a model hash, then certify it to receive an X.509 certificate and a certification status. Events streamed to Kakunin are used for monitoring and reporting.

What happens when an agent misbehaves or crosses a risk threshold?

Behavioral monitoring drives risk scoring over a rolling window, and risk-threshold behavior can trigger auto-revocation/blocks and webhook alerts, along with recorded audit events for traceability.

Where are private keys stored?

Private key custody is handled via AWS KMS only, aligning certificate-based cryptographic operations with controlled key management.

How can I integrate Kakunin into my application?

Use Kakunin’s REST API or the TypeScript SDK. HMAC-signed webhooks and an OpenAPI 3.0 spec are provided, and sandbox mode supports testing in a sandbox environment.

Does Kakunin replace human KYC or model governance tools?

No—Kakunin is complementary. It focuses on cryptographic identity and behavioral accountability for AI agents, alongside human KYC and model governance practices.