Architecture

Architecture

mysql-mcp provides a robust, modular foundation designed for scalability, zero-trust telemetry, and secure AI agent orchestration. Read the full value proposition.

Unlock Core Capabilities

• Specialized Tool Library distributed across Tool Groups to handle CRUD and spatial analysis.
• Shortcuts (Tool Filtering) to configure context-friendly setups.
• Observability Resources and AI-Powered Prompts to support agent workflows.

Scale Infrastructure Seamlessly

• Code Mode (mysql_execute_code): Advanced sandbox for multi-step data manipulation executing securely within a native isolated-vm V8 instance inside the MCP server without round trips. This delivers cost savings and reduces error points. It includes configurable payload caps and distributed Redis-backed rate limiting (with seamless in-memory fallback).
• Supported Transports: Switch between stdio, http (Streamable HTTP /mcp), and sse (Legacy /sse).
• Observability & Telemetry: Native Prometheus /metrics export on HTTP transport for pre-configured Grafana dashboards.
• Authentication: We support Simple Bearer Tokens or full OAuth (RFC 9728/8414) via OIDC-compliant Enterprise Identity Providers.

Review Architectural Pillars

• Strict Execution Boundaries: Secure parameters and query aliases at the MCP layer before interacting with the database.
• Optimized Payload Delivery: Require explicit filters for high-cardinality operations to guarantee token efficiency.

Enforce Robust Security

• Dual-Schema Validation Pattern: Tools use a Base schema for visibility. They use a Derived schema for parsing. This supports aliases without breaking JSON Schema. Integrates Standard Schema interfaces via Zod's safe parsing patterns to enforce tool boundaries. This rejects invalid or hallucinated payloads at the MCP boundary, ensuring predictable AI agent execution.