🆕 Rules Version 2026-Q1.1 · EU AI Act Ready

Scan for EU AI Act compliance in under 3ms.

Zero egress. 100% offline WASM engine. Block non-compliant AI requests at the edge — before they consume a single LLM token.

npx @radu_api/sentinel-scan ./manifest.json

╔══════════════════════════════════════════╗

║ 🛡 SENTINEL SCAN — EU AI Act Auditor ║

╚══════════════════════════════════════════╝

─────────────────────────────────────

App: dark-social-scorer @ v1.0.0

Rules: 2026-Q1.1

─────────────────────────────────────

Verdict: ❌ NON_COMPLIANT

✗ [ART5-003] FORBIDDEN: social_scoring

Everything your AI team needs for compliance.

⚡

WebAssembly Core

The EU AI Act ruleset is compiled into a 40KB WASM binary. Runs at native speed — no interpreter, no cold starts.

🔒

Art. 5 Hard Block

Forbidden practices (social scoring, subliminal manipulation, biometric surveillance) are rejected instantly, before any AI call.

🌐

Edge API (500+ PoPs)

Deploy your compliance gate globally on Cloudflare's network. Block non-compliant requests in the same region as your users.

🤖

AI Fallback (Llama 3)

Ambiguous cases escalate to Llama 3 with a 90% confidence guard. No hallucinations. No verdict without evidence.

📋

Full Audit Trail

Every verdict is logged in D1 with client_id, rules_version, and timestamp. Ready for your next conformity assessment.

⚙️

CI/CD Native

Exit code 0 = COMPLIANT. Exit code 1 = NON_COMPLIANT. Drop npx @radu_api/sentinel-scan into any pipeline.

One plan. Full access. Cancel anytime.

Starter

$99

per month · billed monthly