MCP/MCP Production Kit

MCP Production Kit

Production-grade, type-safe MCP server boilerplate with dual transports (stdio & SSE), SSRF guards, agent memory KV, and zero-leak logging.

developer-tools•Developer Tooling & Discovery
GitHub15
License:MIT
VERIFIED
ByAhmed@Ahmed

About this Server

MCP Production Kit is an enterprise-ready, type-safe Model Context Protocol (MCP) server starter boilerplate. It features a dual-transport architecture supporting local stdio and remote HTTP/Server-Sent Events (SSE) from a single codebase, zero-leak stderr protocol logging, SSRF-hardened HTTP fetching, namespaced in-memory KV state with TTL expiration, and system telemetry.

Real-time Interaction

Direct low-latency RPC protocol communication with structured JSON schemas.

Scoped Sandboxing

Restricted capability bounds to prevent accidental modifications or leakage.

Security & Best Practices

Zero-Leak Stderr Logging: Exclusively reserves stdout for JSON-RPC 2.0 framing while piping diagnostic traces to stderr with credential redaction.
SSRF Network Guard: Rejects loopback addresses, private IP ranges, and cloud instance metadata APIs by default.
Strict Memory Bounds: Limits payload responses to 512KB and enforces 10-second fetch timeouts to prevent memory exhaustion.

When to Use (and When Not To)

Best For
  • •Bootstrapping new production-ready MCP servers with dual stdio and SSE transport needs
  • •Providing secure, SSRF-protected HTTP scraping and API exploration capabilities to AI assistants
  • •Persisting session-scoped agent variables and intermediate workflow steps with TTL cleanup
Not For / Limitations
  • •Kubernetes cluster pod troubleshooting and kubectl rollout scaling (use Kubernetes MCP Server instead)
  • •Agent rules and markdown steering file linting (use Asitis Agent Tools instead)
  • •Relational database point-in-time recovery and snapshot rollbacks (use Rowsafe MCP instead)

Install & Setup

terminal command
$claude mcp add production-kit -- node /path/to/mcp-production-kit/dist/index.js --transport=stdio

Exposed Tools & Endpoints

3 tools

Functions your AI agent can invoke via this Model Context Protocol server

secure_http_fetch

Performs SSRF-hardened web fetching blocking private IPs (10.0.0.0/8, 192.168.0.0/16, AWS metadata 169.254.169.254) with a 512KB payload ceiling.

agent_memory_store

Provides namespaced key-value state persistence (set, get, list, delete) with configurable automatic TTL expiration.

get_system_metrics

Inspects real-time OS host telemetry, CPU load averages, and NodeJS heap memory consumption.