Security
Guardrails is built with security as a first-class concern, both in runtime enforcement and platform design.
API key security
API keys grant access to validation, profiles, and analytics.
Best practices
- Never expose API keys in client-side code
- Rotate keys regularly
- Use least-privilege keys
- Revoke unused keys immediately
Transport security
All communication with Guardrails uses:
- HTTPS only
- TLS 1.2+
- Encrypted payloads in transit
Requests over plain HTTP are rejected.
Data handling
Guardrails is designed to minimize sensitive data exposure:
- Input text is processed in memory
- Analytics payloads are structured and scoped
- Optional redaction guardrails can be enabled
- Retention policies are configurable
Access control
Guardrails supports:
- Per-user isolation
- Per-API-key limits
- Profile-based enforcement
- Tool-level permissions
Auditing and observability
Every execution can emit structured analytics events:
- Guardrail pass/fail
- Execution latency
- Profile usage
- Rate limit hits
These events enable:
- Compliance audits
- Security monitoring
- Incident investigation
Compliance readiness
Guardrails is designed to support regulated environments such as:
- Healthcare (HIPAA)
- Finance
- Education
- Enterprise SaaS
Recommended guardrails
For high-security deployments, enable:
- Secrets detection
- PII redaction
- Internal data leak prevention
- Prompt injection protection
- Tool access control
Next steps
- Configure environment variables → Environment Setup
- Prepare for high traffic → Scaling