PromptKit MCP Server

AI Prompt Library & Engineering System - MCP Server Implementation

🚀 Quick Start

$3

``bash


From the monorepo root

pnpm install

Build PromptKit

cd implementations/mcp-servers/prompt-kit
pnpm install
pnpm build
`

$3

Add to your Claude Code config (~/.config/claude/config.json):

`json
{
"mcpServers": {
"prompt-kit": {
"command": "node",
"args": ["/path/to/fast-kit/implementations/mcp-servers/prompt-kit/dist/index.js"]
}
}
}
`

🎯 Features

- ✅ 23 curated prompt templates across 6 categories
- ✅ Template composition with Handlebars
- ✅ Variable validation
- ✅ Usage analytics tracking
- ✅ Custom prompt creation
- ✅ Semantic search
- ⏳ More prompts coming (target: 100+)
- ⏳ Prompt chaining (coming soon)

📦 Available Tools

$3

List available prompt templates with filtering.

Parameters:
- category (optional): Filter by category
- tags (optional): Filter by tags
- search (optional): Search query
- limit (optional): Max results

$3

Get a specific prompt template.

Parameters:
- prompt_id (required): Prompt template ID
- include_examples (optional): Include usage examples

$3

Compose a prompt from template with variables.

Parameters:
- prompt_id (required): Prompt template ID
- variables (required): Variables object
- inject_context (optional): Auto-inject context

Example:
`json
{
"prompt_id": "function_creation",
"variables": {
"function_name": "calculateTax",
"purpose": "Calculate tax amount based on price and tax rate",
"language": "typescript",
"parameters": [
{"name": "price", "type": "number", "description": "Base price"},
{"name": "taxRate", "type": "number", "description": "Tax rate (0-1)"}
],
"return_type": "number"
}
}
`

$3

Search for prompts semantically.

Parameters:
- query (required): Search query
- limit (optional): Max results

$3

Create a custom prompt template.

Parameters:
- name (required): Prompt name
- category (required): Category
- description (required): Description
- template (required): Handlebars template
- variables (optional): Variable definitions

$3

Track prompt usage for analytics.

Parameters:
- prompt_id (required): Prompt ID
- success (required): Was it successful?
- feedback (optional): User feedback

📚 Built-in Prompt Library

$3

- function_creation - Generate function implementations
- api_endpoint - Create REST/GraphQL API endpoints
- react_component - Generate React components
- database_schema - Design database schemas
- algorithm_implementation - Implement algorithms with optimal complexity

$3

- unit_test_creation - Generate comprehensive unit tests
- integration_test - Create integration tests
- e2e_test - End-to-end test generation
- test_fixtures - Create test data and fixtures

$3

- fix_bug - Debug and fix bugs with analysis
- analyze_performance - Debug performance bottlenecks
- debug_async_issue - Fix async/concurrency issues

$3

- refactor_code - Refactor for quality improvements
- extract_function - Extract code to functions (DRY)
- simplify_logic - Simplify complex conditional logic
- modernize_code - Update to modern language features

$3

- review_pull_request - Comprehensive PR review
- security_audit - Security vulnerability scan
- performance_review - Performance optimization review

$3

- add_documentation - Generate code documentation

$3

- system_design - High-level system architecture
- api_design - Comprehensive API design
- design_patterns - Apply appropriate design patterns

Total: 23 high-quality templates ready to use!

💾 Data Storage

`
~/.fast-kit/
├── prompts/
│ └── custom/
│ └── {prompt-id}.yaml
└── analytics/
└── prompt-usage.db
`

🎨 Prompt Template Format

`yaml
id: my_prompt
category: code_generation
name: My Custom Prompt
description: What this prompt does
version: 1.0.0

metadata:
author: your-name
created_at: '2026-01-10'
tags: [tag1, tag2]

variables:
- name: var_name
type: string
description: Variable description
required: true

template: |
# Your prompt template using {{var_name}}

## Instructions
...

settings:
temperature: 0.3
max_tokens: 2000
`

📖 Examples

$3

`bash


Using Claude Code


> Use the function_creation prompt to create a validateEmail function in TypeScript
`

Claude will call compose_prompt and generate a detailed prompt for implementing the function.

$3

`bash
> Use the unit_test_creation prompt to write tests for my calculateDiscount function
`

$3

`bash
> Search for prompts about "testing APIs"
`

$3

`bash
> Create a custom prompt for generating React components
`

🧪 Analytics

PromptKit tracks:
- Total uses per prompt
- Success rate
- Average tokens
- User feedback

View stats:
`bash


Get prompt with stats


> Show me stats for the function_creation prompt
`

🔧 Development

$3

1. Create a YAML file in prompts/builtin/{category}/
2. Follow the template format
3. Test with compose_prompt

$3

`bash
pnpm test
``

📝 License

MIT