open-guardrailLLM Guardrail Engine

350 guards, 26 PII regions, 8 adapters, <0.1ms latency. Guard LLM input/output in one line of code.

🛡️

350 Built-in Guards

Security, privacy (26 PII regions), compliance (12 regulations), content safety (13 profanity languages), agent safety — ready out of the box.

⚡

Single guard runs at 1M+ ops/s. 6-guard pipeline under 0.1ms. No external API calls needed.

🔌

OpenAI, Anthropic, Next.js, Express, Fastify, Hono, Vercel AI SDK, LangChain.

📄

Define guards in YAML/JSON. No code changes needed. Hot-reload friendly.

🌊

Chunk-level validation with StreamingPipeline. Guard text as it streams.

🇰🇷

GDPR, EU AI Act, Korean AI Basic Act, ISMS-P, PIPA, PCI DSS, SOX, FERPA, HIPAA, COPPA — 12 regulations built-in.