Architecture
Value Proposition Build upon a robust, modular foundation designed for scalability, zero-trust telemetry, and secure AI agent orchestration. Read the full value proposition.
- Comprehensive Specialized Tool Library engineered to seamlessly orchestrate complex database operations—ranging from essential CRUD tasks to advanced spatial analysis—while ensuring high reliability and maximum token efficiency.
- Shortcuts (Tool Filtering) to configure context-friendly setups.
- Observability Resources and AI-Powered Prompts to support agent workflows.
-
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(unified modern HTTP/SSE), andsse(Legacy/sse). -
Observability & Telemetry: Native Prometheus
/metricsexport on HTTP transport for pre-configured Grafana dashboards. - Authentication: We support Simple Bearer Tokens or full OAuth 2.1 authentication via OIDC-compliant Enterprise Identity Providers.
- Strict Execution Boundaries: Enforce zero-trust security by validating parameters and query aliases at the MCP layer prior to database execution.
- Optimized Payload Delivery: Drastically reduce LLM token costs by enforcing explicit filters for high-cardinality operations.
-
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 strictly typed Zod boundaries (
.safeParse()). This provides deterministic, type-safe I/O boundaries that gracefully reject hallucinated AI payloads without throwing unhandled exceptions.