MCP Toolkit

MCP server with evaluation harness, prompt library, and deployment. The standard stack for building tool servers for AI systems.

Name	`mcp-toolkit`
Version	`0.1.0`
Tags	mcp, tools, ai, typescript
Source	`composites/mcp-toolkit.yaml`

Render it

# CLI
npx @nanohype/sdk render-composite mcp-toolkit --out ./my-app

# SDK
import { LocalSource, renderComposite } from "@nanohype/sdk";
const result = await renderComposite(source, "mcp-toolkit", variables);

# MCP — from an agent
get_composite({ name: "mcp-toolkit" })

What it draws together

Template	Rendered at	Notes
`mcp-server-ts`	root	the root project
`prompt-library`	`prompts`	only when `IncludePrompts`
`eval-harness`	`evals`	only when `IncludeEvals`

Variables

Required

ProjectName string: Project name used across all templates Must be lowercase kebab-case.
ServerName string: Human-readable MCP server name

Optional

IncludePrompts bool — defaults to true: Include versioned prompt library
IncludeEvals bool — defaults to true: Include evaluation harness