This is a personal proof-of-concept project. It is not intended for production use. Please refrain from using it.

Aira Documentation

The authorization and audit layer for AI agents. Gate every action, sign every outcome, prove to any regulator.

Getting started

Gateway — zero codeRoute any LLM call through Aira. Every request scanned, policy-checked, and receipted.QuickstartAuthorize your first action in 5 minutes with the SDK.Python SDKpip install aira-sdk TypeScript SDKnpm install aira-sdk AuthenticationAPI keys, JWT, OAuth.

Platform

Policy engine4 modes: rules, AI, consensus, content scan Content scannerNER (Presidio) + 27 regex patterns LLM GatewayAny provider, zero code change Human approvalSigned decisions in the receipt chain Ed25519 receiptsEvery state signed, RFC 3161 timestamps W3C DIDVerifiable agent identity Output filteringFlag, deny, or redact LLM responses Multi-party signingTwo independent keys per receipt Drift detectionBehavioral baselines + KL divergence Merkle settlementTamper-evident batch anchoring Self-hostedYour Kubernetes, your keys, your data Consensus scoringMulti-model voting, deterministic

Compliance

EU AI ActArticles 6, 9, 12, 14, Annex IV DORAArticles 17-19, 24-27, 28-44 DORA incidentsMajor-incident PDFs, ESA submission DORA third partiesICT vendor register Compliance bundlesMerkle-rooted evidence exports Compliance reportsRegulator-ready signed PDFs

API Reference

Actions Policies Gateway Cases Receipts Verification Agents Evidence DORA Webhooks Billing All endpoints →

EU AI Act · DORA · SR 11-7 · ISO 42001 · SOC 2 CC7 · GDPR