Provides pre-defined prompt templates for AI assistants to generate comprehensive plans for TypeScript projects, API architectures, and GitHub workflows.
A Model Context Protocol (MCP) server that provides pre-defined prompt templates for AI assistants, allowing them to generate comprehensive plans for TypeScript projects, API architectures, and GitHub workflows.
This MCP server provides a set of prompt templates that can be used by AI assistants to generate detailed, structured responses for TypeScript project planning. It offers templates for: - Creating comprehensive API architecture plans - Setting up new TypeScript projects with best practices - Generating GitHub workflow configurations
This MCP was specifically created to work with the Local Dev MCP, forming a powerful combination where the Prompt MCP generates detailed project plans and the Local Dev MCP executes them. Together, they create a seamless workflow for AI-assisted TypeScript project development.
Each prompt template is designed to ensure AI assistants provide consistent, high-quality, and detailed project plans following modern TypeScript development standards.
Clone the repository
git clone <repository-url>
cd typescript-prompt-mcpInstall dependencies
npm installSet up environment variables
# Create development environment file
cp .env.example .env.development
# Create production environment file
cp .env.example .env.productionnpm run devThis starts the MCP server in development mode with hot reload.
npm run build
npm startOr use the shorthand:
npm run prodTo add this MCP server to Claude Desktop:
Start the MCP server Make sure your server is running locally.
Open Claude Desktop settings
Select "Settings" from the dropdown menu
Navigate to Extensions settings
4.1 Configure the MCP connection
- Name: TypeScript Prompt MCP (or any name you prefer)
- URL: Enter the URL where your MCP server is running (e.g., http://localhost:3000 for local development)
- Click "Add MCP"
4.2 Alternative: Configure the MCP connection via command - You first need to build the project and provide your full path to the compiled server - Add the following to your Claude Desktop configuration:
"ts-prompts": {
"command": "node",
"args": [
"YOUR_CUSTOM_PATH/dist/index.js"
]
}Claude Desktop will attempt to connect to your MCP server
Add Local Dev MCP
Having both MCPs enabled allows for a complete workflow from planning to implementation
Verify connection
Once connected with both MCPs, you can ask Claude to:
This combination of MCPs creates a powerful workflow where you can plan your project in detail and then implement it without leaving the Claude interface.
The server exposes several prompts that can be used by AI assistants:
api-architectureGenerates a comprehensive architecture plan for a TypeScript API.
Parameters:
- projectName: Name of the API project
- database: Database to use (postgres, mysql, mongodb, etc.)
- auth: Authentication method (jwt, oauth, none)
- endpoints: Comma-separated list of main API endpoints
new-project-setupGenerates a comprehensive setup plan for a new TypeScript project.
Parameters:
- projectName: Name of the project
- projectType: Type of project (api, frontend, library, cli)
- features: Key features or requirements separated by commas
github-workflowGenerates a GitHub workflow plan for a TypeScript project.
Parameters:
- projectName: Name of the project
- ciFeatures: Comma-separated list of CI features (lint, test, build, etc.)
- deployTarget: Deployment target (netlify, vercel, aws, azure, etc.)
- branchStrategy: Branch strategy (gitflow, trunk, github-flow)
The server creates an MCP server using the ModelContextProtocol SDK:
src/
├── index.ts # Entry point that sets up the MCP server
├── prompts/ # Prompt definitions
│ ├── apiArchitecture.ts # API architecture prompt
│ ├── githubWorkflow.ts # GitHub workflow prompt
│ ├── newProjectSetup.ts # New project setup prompt
│ └── index.ts # Exports all prompts
scripts/
├── prepare-build.ts # Script for preparing production builds
├── run-relevant-tests.ts # Script for running tests on changed files
└── setup-husky.js # Script for setting up Git hooksTo add a new prompt template:
src/prompts directorymcpServer.prompt() methodsrc/prompts/index.tsExample:
import { z } from 'zod';
import { mcpServer } from '../index';
mcpServer.prompt(
'my-new-prompt',
'Description of what this prompt does',
{
param1: z.string().describe('Description of param1'),
param2: z.number().optional().describe('Description of param2'),
},
async ({ param1, param2 = 0 }) => {
return {
messages: [
{
role: 'user',
content: {
type: 'text',
text: `Your prompt template with ${param1} and ${param2}...`,
},
},
],
};
},
);The server uses different environment files for development and production:
- .env.development - Used when running in development mode
- .env.production - Used when running in production mode
Run the test suite with:
npm test# Run ESLint
npm run lint
# Fix ESLint errors
npm run lint:fix
# Format code with Prettier
npm run format
# Check formatting
npm run format:checkWhen deploying to production:
.env.production file contains valid credentials if requirednpm run prod to build and start the production serverThis project is licensed under the MIT License - see the LICENSE file for details.
Gpaul | Faldin