Pre-execution policy enforcement
Vaikora evaluates every proposed AI agent action against a deterministic policy engine before the call leaves the perimeter. Decisions land in under 500ms at the p95. The engine returns one of four verdicts: ALLOW, ALLOW_LOG, CONSTRAIN, or BLOCK. High-risk actions can also be routed to a human approver in Slack with the agent held in a WAIT state until resolution.
Cryptographic audit chain
Every enforcement decision is signed into a SHA-256 append-only audit chain. Auditors can replay the chain end-to-end and verify that no record was modified after the fact. Receipts include the action payload, the matched policy, the decision, the agent, and the timestamp.
Pre-built compliance presets
Vaikora ships configuration presets for the five most common enterprise compliance frameworks: SOC 2 Type II, HIPAA, GDPR, PCI DSS, and ISO 27001. The presets cover logging, evidence collection, and policy-mapping requirements out of the box, so audit prep happens at deploy time, not under deadline pressure.
Open core
The Vaikora reference gateway is MIT-licensed and free forever. The companion MCP server ships in the Anthropic MCP registry and is installable from PyPI. The commercial Vaikora Control Plane adds the audit chain, compliance presets, approvals UI, and SLA. Buy the control plane or self-host the gateway.
Built-in content modules
The Vaikora gateway includes six content modules out of the box: PII detection, jailbreak detection, prompt injection detection, semantic risk classification, domain risk scoring, and email classification. Each module can be enabled, configured, or replaced independently. Each runs synchronously inside the policy pipeline so decisions stay deterministic.
Distribution channels
Vaikora is available through AWS Marketplace, Azure Sentinel, the Anthropic MCP registry, and direct API. Procurement teams in regulated industries can purchase Vaikora through whichever path matches their existing cloud commitment.