Skip to content

Tae4an/mcp-prompt-manager

BranchesTags

Repository files navigation

MCP Prompt Manager

Certified by MCP Review

This project is officially certified by MCP Review.

MCP (Model Context Protocol) Prompt Manager is a server that enables AI models like Claude to access local prompt files. It provides functionality for creating, retrieving, updating, and deleting prompts, allowing efficient management of frequently used prompts.

Key Features

Core Features

  • List all prompts
  • Retrieve specific prompt content
  • Create new prompts
  • Update prompt content
  • Delete prompts

Advanced Features

  • Intelligent Search: Advanced fuzzy search with multiple algorithms (Levenshtein, Jaro-Winkler, n-gram similarity)
  • Category & Tag System: Organize prompts with categories and tags
  • Template Processing: Use variable substitution with {{variable}} syntax and advanced conditionals
  • Template Library: Built-in template library with 12 professional templates across 5 categories
  • Favorites Management: Mark frequently used prompts as favorites
  • Metadata Management: Automatic metadata tracking for enhanced organization
  • Version Management: Complete version control with history tracking, diff comparison, and rollback capabilities
  • Import/Export System: Backup and restore prompts with JSON format, including metadata and version history
  • Security & Validation: Comprehensive input sanitization, rate limiting, and error handling
  • Caching System: Intelligent caching for improved performance
  • Structured Logging: Advanced logging with multiple levels and file output

Installation

Prerequisites

  • Node.js v18 or higher
  • npm

Installation Steps

  1. Clone the repository

    git clone https://github.com/Tae4an/mcp-prompt-manager.git
cd mcp-prompt-manager

  2. Install dependencies

    npm install

  3. Grant execution permissions

    chmod +x server.js

Containerized Run

Run with Docker

docker build -t mcp-prompt-manager:local .
docker run --rm \
  -e NODE_ENV=production \
  -e LOG_DIR=/var/log/mcp \
  -e PROMPTS_DIR=/data/prompts \
  -v $(pwd)/prompts:/data/prompts \
  -v $(pwd)/logs:/var/log/mcp \
  mcp-prompt-manager:local

Run with Docker Compose

docker compose up -d --build

Environment variables

  • PROMPTS_DIR: Directory to store prompts (default: project prompts)
  • LOG_DIR: Directory for file logging (default: ./logs)
  • Cache TTL/size:
    • FILE_CACHE_TTL, FILE_CACHE_MAX_SIZE
    • SEARCH_CACHE_TTL, SEARCH_CACHE_MAX_SIZE
    • METADATA_CACHE_TTL, METADATA_CACHE_MAX_SIZE
    • TEMPLATE_CACHE_TTL, TEMPLATE_CACHE_MAX_SIZE
  • Rate limit presets:
    • RATE_LIMIT_STANDARD_WINDOW_MS, RATE_LIMIT_STANDARD_MAX
    • RATE_LIMIT_STRICT_WINDOW_MS, RATE_LIMIT_STRICT_MAX
    • RATE_LIMIT_LENIENT_WINDOW_MS, RATE_LIMIT_LENIENT_MAX
    • RATE_LIMIT_UPLOAD_WINDOW_MS, RATE_LIMIT_UPLOAD_MAX
  • Policy/permissions:
    • READ_ONLY (true/false): disable write operations
    • DISABLE_IMPORT (true/false): block import
    • DISABLE_EXPORT (true/false): block export
    • DISABLE_VERSION_ROLLBACK (true/false): block rollback

Connecting to Claude Desktop

  1. Install Claude Desktop (if not already installed)

  2. Open Claude Desktop configuration file:

    • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
    • Windows: %APPDATA%\Claude\claude_desktop_config.json

  3. Add the following content to the configuration file:

    {
  "mcpServers": {
    "promptManager": {
      "command": "node",
      "args": ["absolute_path_to_cloned_repository/server.js"]
    }
  }
}

    Example:

    {
  "mcpServers": {
    "promptManager": {
      "command": "node",
      "args": ["/Users/username/projects/mcp-prompt-manager/server.js"]
    }
  }
}

  4. Restart Claude Desktop

Usage

In Claude Desktop, click the tools icon (🛠️) to access the following MCP tools:

Core Tools

list-prompts

Retrieves a list of all prompts.

  • Parameters: None

get-prompt

Retrieves the content of a specific prompt.

  • Parameters: filename - Name of the prompt file to retrieve

create-prompt

Creates a new prompt.

  • Parameters:
    • filename - Name of the prompt file to create (e.g., my-prompt.txt)
    • content - Prompt content

update-prompt

Updates the content of an existing prompt.

  • Parameters:
    • filename - Name of the prompt file to update
    • content - New prompt content

delete-prompt

Deletes a prompt (automatically removes associated metadata).

  • Parameters: filename - Name of the prompt file to delete

Advanced Tools

search-prompts

Advanced fuzzy search prompts by filename or content with intelligent ranking.

  • Parameters:
    • query - Search query string
    • searchInContent - (Optional) Boolean to search within prompt content (default: false)
    • limit - (Optional) Maximum number of results to return (default: 10)
    • threshold - (Optional) Minimum similarity threshold (0.0-1.0, default: 0.3)

tag-prompt

Add tags to a prompt for better organization.

  • Parameters:
    • filename - Name of the prompt file to tag
    • tags - Array of tag strings

categorize-prompt

Set a category for a prompt.

  • Parameters:
    • filename - Name of the prompt file to categorize
    • category - Category name string

list-by-category

List prompts organized by category.

  • Parameters:
    • category - (Optional) Specific category to filter by

process-template

Process a prompt template with variable substitution.

  • Parameters:
    • filename - Name of the template prompt file
    • variables - Object with variable names as keys and replacement values as values
  • Note: Use {{variable}} format in templates

list-template-variables

List all variables found in a template prompt.

  • Parameters:
    • filename - Name of the template prompt file to analyze

favorite-prompt

Add or remove a prompt from favorites.

  • Parameters:
    • filename - Name of the prompt file
    • action - Either "add" or "remove"

list-favorites

List all favorite prompts with detailed information.

  • Parameters: None

Version Management Tools

list-prompt-versions

List all versions of a specific prompt with timestamps and actions.

  • Parameters:
    • filename - Name of the prompt file to get version history for

compare-prompt-versions

Compare two versions of a prompt and show detailed differences.

  • Parameters:
    • filename - Name of the prompt file to compare
    • fromVersion - Source version number to compare from
    • toVersion - Target version number to compare to

rollback-prompt

Rollback a prompt to a specific previous version.

  • Parameters:
    • filename - Name of the prompt file to rollback
    • version - Version number to rollback to

get-prompt-version

Get the content of a specific version of a prompt.

  • Parameters:
    • filename - Name of the prompt file
    • version - Version number to retrieve

get-prompt-version-stats

Get statistics about a prompt's version history including total versions, actions breakdown, and size history.

  • Parameters:
    • filename - Name of the prompt file to get statistics for

Template Library Tools

The built-in template library includes 12 professional templates across 5 categories:

Available Template Categories:

  • 🖥️ Coding & Development (3 templates): Code review, debugging help, API documentation
  • 🌐 Translation & Language (2 templates): Text translation, grammar checking
  • 📝 Document Writing (2 templates): Document summarization, meeting minutes
  • 📊 Analysis & Research (2 templates): SWOT analysis, competitive analysis
  • 🎓 Education & Learning (3 templates): Lesson plans, quiz generation

list-template-categories

List all available template categories with descriptions and template counts.

  • Parameters: None

list-templates-by-category

List all templates in a specific category.

  • Parameters:
    • categoryId - Category ID to list templates from

get-template-details

Get detailed information about a specific template including variables and usage.

  • Parameters:
    • templateId - Template ID (format: category.template-name)

search-templates

Search through the template library using fuzzy matching.

  • Parameters:
    • query - Search query string
    • category - (Optional) Filter by specific category
    • tags - (Optional) Array of tags to filter by
    • limit - (Optional) Maximum number of results (default: 10)

render-template

Render a template with provided variables and get the processed content.

  • Parameters:
    • templateId - Template ID to render
    • variables - Object with variable names and values
    • sanitizeOutput - (Optional) Enable output sanitization (default: true)

validate-template

Validate template syntax and check for potential issues.

  • Parameters:
    • templateId - Template ID to validate

get-popular-templates

Get list of most popular templates based on usage patterns.

  • Parameters:
    • limit - (Optional) Number of templates to return (default: 5)

get-related-templates

Get templates related to a specific template based on tags and categories.

  • Parameters:
    • templateId - Template ID to find related templates for
    • limit - (Optional) Number of related templates (default: 3)

get-template-library-stats

Get comprehensive statistics about the template library.

  • Parameters: None

create-prompt-from-template

Create a new prompt file using a template with variable substitution.

  • Parameters:
    • templateId - Template ID to use
    • filename - Name for the new prompt file
    • variables - Object with template variables
    • addMetadata - (Optional) Add template metadata to file (default: true)

Import/Export Tools

export-prompts

Export prompts to JSON format for backup or sharing.

  • Parameters:
    • format - (Optional) Export format: "json" (default: json)
    • includeMetadata - (Optional) Include metadata in export (default: true)
    • includeVersionHistory - (Optional) Include version history (default: false)
    • filterByTags - (Optional) Array of tags to filter prompts
    • filterByCategory - (Optional) Category to filter prompts
    • compress - (Optional) Compress export data (default: false)

import-prompts

Import prompts from JSON format with validation and conflict resolution.

  • Parameters:
    • importData - Import data object in export format
    • overwriteExisting - (Optional) Overwrite existing files (default: false)
    • skipDuplicates - (Optional) Skip duplicate files (default: true)
    • validateChecksums - (Optional) Validate file checksums (default: true)
    • createBackup - (Optional) Create backup before import (default: true)
    • mergeMetadata - (Optional) Merge with existing metadata (default: true)

get-import-export-status

Get import/export system status and capabilities.

  • Parameters: None

Operations/Monitoring Tools

get-server-stats

Get process/server runtime stats (uptime, memory, policy flags, cache info).

  • Parameters: None

get-policy-status

Get current policy/permission flags derived from environment variables.

  • Parameters: None

get-cache-stats

Read cache statistics for files/metadata/search/templates (size, hit/miss, memory usage).

  • Parameters: None

get-rate-limit-status

Get current rate limiter stats for standard/strict/upload presets.

  • Parameters: None

Technical Features

Security & Performance

  • Input Sanitization: Comprehensive XSS and injection attack prevention
  • Rate Limiting: Configurable rate limiting with sliding window algorithm
  • Caching System: Multi-level LRU caching with TTL support for improved performance
  • Error Handling: Advanced error recovery and logging system
  • File Validation: SHA-256 checksums and integrity verification

Advanced Template Engine

  • Conditional Logic: Support for {{#if}}, {{#unless}}, {{#each}} constructs
  • Loop Processing: Iterate over arrays and objects in templates
  • Function Calls: Built-in helper functions for formatting and processing
  • Nested Variables: Support for complex object structures
  • Error Recovery: Graceful handling of missing variables and malformed templates

Fuzzy Search Algorithms

  • Levenshtein Distance: Character-based similarity matching
  • Jaro-Winkler Distance: Optimized for prefix matching
  • N-gram Similarity: Substring pattern matching
  • Intelligent Ranking: Multi-factor scoring with customizable thresholds
  • Highlighting: Search result highlighting for better user experience

Data Management

  • Version Control: Complete history tracking with diff comparison
  • Metadata System: Automatic tagging, categorization, and favorites
  • Backup System: Automated backup creation during import operations
  • Export Formats: JSON with optional compression and filtering
  • File Organization: Structured storage with hidden metadata directories

Advanced Configuration

Changing Prompt Storage Path

By default, prompts are stored in the prompts folder in the directory where the server file is located. You can change the path using environment variables:

PROMPTS_DIR=/desired/path node server.js

Or set environment variables in claude_desktop_config.json:

{
  "mcpServers": {
    "promptManager": {
      "command": "node",
      "args": ["/absolute/path/mcp-prompt-manager/server.js"],
      "env": {
        "PROMPTS_DIR": "/desired/path"
      }
    }
  }
}

Examples

Basic Usage

  1. Creating a new prompt:

    • Tool: create-prompt
    • Filename: greeting.txt
    • Content: You are a friendly and helpful AI assistant. Please respond politely to user questions.

  2. Listing prompts:

    • Tool: list-prompts

  3. Retrieving prompt content:

    • Tool: get-prompt
    • Filename: greeting.txt

Advanced Usage

  1. Creating a template prompt:

    • Tool: create-prompt
    • Filename: email-template.txt
    • Content: Dear {{name}}, Thank you for your interest in {{product}}. Best regards, {{sender}}

  2. Processing a template:

    • Tool: process-template
    • Filename: email-template.txt
    • Variables: {"name": "John", "product": "MCP Server", "sender": "Support Team"}

  3. Organizing prompts:

    • Tool: categorize-prompt

    • Filename: greeting.txt

    • Category: customer-service

    • Tool: tag-prompt

    • Filename: greeting.txt

    • Tags: ["polite", "professional", "greeting"]

  4. Searching prompts:

    • Tool: search-prompts
    • Query: assistant
    • SearchInContent: true

  5. Managing favorites:

    • Tool: favorite-prompt
    • Filename: greeting.txt
    • Action: add

Version Management Usage

  1. Viewing version history:

    • Tool: list-prompt-versions
    • Filename: greeting.txt

  2. Comparing versions:

    • Tool: compare-prompt-versions
    • Filename: greeting.txt
    • FromVersion: 1
    • ToVersion: 3

  3. Rolling back to a previous version:

    • Tool: rollback-prompt
    • Filename: greeting.txt
    • Version: 2

  4. Getting version statistics:

    • Tool: get-prompt-version-stats
    • Filename: greeting.txt

Template Library Usage

  1. Browsing template categories:

    • Tool: list-template-categories

  2. Using a template:

    • Tool: render-template
    • TemplateId: coding.code-review
    • Variables: {"code": "function hello() { console.log('Hello'); }", "language": "javascript"}

  3. Creating prompt from template:

    • Tool: create-prompt-from-template
    • TemplateId: writing.meeting-minutes
    • Filename: weekly-standup.txt
    • Variables: {"meeting_title": "Weekly Standup", "date": "2024-08-04", "attendees": "Team Alpha"}

  4. Searching templates:

    • Tool: search-templates
    • Query: code review
    • Category: coding

Import/Export Usage

  1. Exporting prompts for backup:

    • Tool: export-prompts
    • IncludeMetadata: true
    • IncludeVersionHistory: false
    • FilterByTags: ["important", "production"]

  2. Importing prompts from backup:

    • Tool: import-prompts
    • ImportData: {exported data object}
    • CreateBackup: true
    • OverwriteExisting: false

  3. Checking import/export status:

    • Tool: get-import-export-status

Advanced Search Usage

  1. Fuzzy search with parameters:
    • Tool: search-prompts
    • Query: custmer servce (intentional typos)
    • SearchInContent: true
    • Threshold: 0.6
    • Limit: 15

Troubleshooting

If the MCP server doesn't connect

  • Verify that the server file path is correct
  • Check that the server has execution permissions
  • Ensure Node.js version is v18 or higher

If tools don't appear

  • Try restarting Claude Desktop
  • Verify that the claude_desktop_config.json file is configured correctly

File access permission issues

  • Ensure you have read/write permissions for the prompts directory

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Support

If you encounter any issues or have questions, please open an issue on the GitHub repository.

About

Local Prompt Management System(Model Context Protocol Server)

Resources

Readme

License

MIT license
Activity

Stars

5 stars

Watchers

1 watching

Forks

3 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages