Production-Grade API Security for Your Perfex CRM

APIs Without Security Are Data Breaches Waiting to Happen

Adding an API to your CRM means exposing your business data to the internet. Client records, financial data, internal metrics — all accessible via HTTP requests.

Most CRM API modules treat security as an afterthought: a single API key, maybe basic rate limiting. That's not enough for production use. Not when your clients' personal data is on the line.

Data Builder takes a different approach. Security isn't a feature we added later. It's the foundation the entire API is built on.

6-Stage Middleware Pipeline

Every API request passes through six security layers before reaching your data. Each layer can reject the request independently:

Stage 1: DDoS Shield

The first line of defense, before any authentication:

• Body size limit: Requests exceeding 256KB are rejected immediately (413)
• Global IP throttle: Per-IP rate limiting prevents brute-force attacks
• Malformed payload detection: JSON nesting deeper than 10 levels or oversized query strings are blocked (400)
• Execution timeout: Requests exceeding 10 seconds are terminated

Stage 2: CORS Guard

Cross-origin request control with two phases:

• Preflight (OPTIONS): Server-level policy — responds without requiring authentication (browsers don't send credentials in preflight)
• Post-auth whitelist: Per-token domain restriction — only allowed origins can make API calls

Stage 3: Auth Gate

Token verification with cryptographic best practices:

• SHA-256 hash-only storage: Plaintext tokens are NEVER stored in the database
• Timing-attack safe comparison: Uses hash_equals() — not ===
• HMAC-SHA256 signing (optional): Verify request integrity with canonical signing (method + path + query + body digest + timestamp)
• Anti-replay protection: X-Timestamp header with 300-second window

Stage 4: Rate Limiter

Per-token request throttling:

• Fixed-window counter with atomic UPSERT (no race conditions)
• Fail-closed: If the rate limit database is unavailable, requests are denied (503) — never allowed through
• Response headers: X-RateLimit-Limit, X-RateLimit-Remaining, Retry-After

Stage 5: Scope Verifier

Fine-grained access control:

• Table-level permissions: Per-table read/create/update/delete matrix
• View-level restrictions: Token can be limited to specific saved views
• CIDR IP matching: Lock tokens to specific IP ranges
• Write enforcement: POST→create, PUT→update, DELETE→delete permission mapping

Stage 6: Request Logger

Complete audit trail:

• Two-phase logging: Start in middleware, finalize after controller (includes status code + response time)
• Sensitive field redaction: token, Authorization, X-Signature are stripped from all stored logs via Log_redactor

OWASP Security Headers

Every API response — success, error, or middleware rejection — includes standard security headers:

These are applied automatically from a single control point — _set_standard_headers(). No configuration needed.

Sensitive Table Protection: 6 Independent Layers

Critical tables (tblstaff, tblcontacts, tbloptions, tbldata_builder_api_tokens) are protected by six independent enforcement mechanisms:

1. Central Registry: ResourceNameResolver::isSensitiveTable() — single source of truth
2. Token Creation: Server-side force create=false, update=false, delete=false for non-Super-Admin
3. Admin Session Bypass: Non-Super-Admin users get read-only synthetic tokens even in API Explorer
4. Scope Verifier Middleware: Runtime 403 on unauthorized write attempts
5. Controller Enforcement: _enforce_write_scope() double-checks before execution
6. GraphQL Schema: Mutation fields are invisible (not just blocked) when token lacks permission

Even if one layer fails, the remaining five still protect your data. Defense in depth.

Sandbox PII Randomization

The API Explorer's sandbox mode includes pattern-based PII randomization for 25+ sensitive column types:

• Email addresses → randomized @example.com
• Phone numbers → randomized format
• Names → synthetic names
• Addresses → randomized locations
• And more...

This applies to both REST sandbox responses and recursive GraphQL tree sanitization — so developers can test query structures without exposing real customer data.

GraphQL-Specific Protections

Zero Compromise Security

Data Builder's security isn't a checkbox feature. It's a production-grade security architecture designed for enterprises that take data protection seriously.

6 middleware stages. OWASP headers. HMAC signing. PII protection. Defense in depth.

Your API is only as strong as its weakest endpoint. We don't have one.

---

Data Builder for Perfex CRM — 6 layers of protection. Zero compromise.