amplify-docs-mcp
This project is a Model Context Protocol server designed to provide powerful and efficient search functionalities for AWS Amplify documentation. It includes features like advanced search syntax, smart results ranking, and automatic updates, all built with TypeScript.
AWS Amplify Documentation MCP Server
A Model Context Protocol (MCP) server that provides powerful search functionality for AWS Amplify documentation. This server clones the official AWS Amplify documentation repository and makes it searchable through a simple MCP tool interface.
Features
- Powerful Search: Search AWS Amplify documentation using natural language queries
- Advanced Search Syntax: Support for boolean operators, wildcards, field-specific search, and more
- Smart Results Ranking: Intelligent ranking of search results based on relevance to query
- Pagination: Navigate through large result sets with pagination support
- Performance Caching: Caching of search results for improved performance
- Auto-Updates: Automatic updates of documentation from the official AWS Amplify repository
- Generation Selection: Choose between Gen 1, Gen 2, or both documentation sets to optimize disk usage and search radius.
- TypeScript Implementation: Built with TypeScript for better type safety and developer experience
Disclaimer: This is a personal project and is not affiliated with, endorsed by, or officially connected to AWS Amplify or Amazon Web Services. This tool is provided as-is without any guarantees or warranty.
Installation
- Clone this repository:
git clone https://github.com/ykethan/amplify-doc-mcp.git
cd amplify-doc-mcp
- Install dependencies:
npm install
- Build the TypeScript code:
npm run build
Configuration
The server is configured using the docs-mcp.config.json
file:
{
"gitUrl": "https://github.com/aws-amplify/docs.git",
"gitRef": "main",
"autoUpdateInterval": 60,
"toolName": "search_amplify_docs",
"toolDescription": "Search AWS Amplify documentation using the probe search engine.",
"ignorePatterns": [
"node_modules",
".git",
"dist",
"build",
"coverage",
".vitepress/cache",
"*.jpg",
"*.jpeg",
"*.png",
"*.gif",
"*.svg",
"*.mp4",
"*.webm"
],
"amplifyGeneration": "gen2"
}
Configuration Options
Option | Description | Default |
---|---|---|
gitUrl | URL of the Git repository to clone for documentation | "https://github.com/aws-amplify/docs.git" |
gitRef | Git branch or tag to checkout | "main" |
autoUpdateInterval | Interval in minutes to check for updates (0 to disable) | 60 |
dataDir | Directory to store documentation data | "./data" |
toolName | Name of the search tool | "search_amplify_docs" |
toolDescription | Description of the search tool | "Search AWS Amplify documentation using the probe search engine." |
ignorePatterns | Array of patterns to ignore when searching | ["node_modules", ".git", ...] |
amplifyGeneration | Which Amplify documentation generation to include | "gen2" |
Auto-Update Mechanism
The server includes an automatic update mechanism that keeps the documentation up-to-date:
- When the server starts, it clones the documentation repository specified in
gitUrl
. - If
autoUpdateInterval
is set to a value greater than 0, the server will periodically check for updates. - Every
autoUpdateInterval
minutes, the server:- Fetches the latest changes from the remote repository
- Checks if the local branch is behind the remote branch
- If updates are available, pulls the changes automatically
- If no updates are needed, continues with the current documentation
This ensures that your documentation search results always include the latest information without requiring a server restart.
Usage
Starting the Server
npm start
Or use the provided start script with options:
./start-server.sh [--gen <1|2|both>] [--rebuild]
Options
--gen 1
: Include only Gen 1 documentation (reduces disk space usage)--gen 2
: Include only Gen 2 documentation (default, reduces disk space usage)--gen both
: Include both Gen 1 and Gen 2 documentation--rebuild
: Force rebuild of data directory
Examples
# Start with Gen 2 documentation (default)
./start-server.sh
# Start with only Gen 1 documentation
./start-server.sh --gen 1
# Start with both Gen 1 and Gen 2 documentation
./start-server.sh --gen both
# Start with only Gen 2 documentation and force rebuild
./start-server.sh --gen 2 --rebuild
Important: When switching between different generation options (e.g., from Gen 1 to Gen 2 or vice versa), it's recommended to use the
--rebuild
flag to ensure a clean repository. This will remove the data directory and clone the repository again, ensuring that the correct files are included based on the selected generation.For example:
# Switch from Gen 2 to Gen 1 with a clean rebuild ./start-server.sh --gen 1 --rebuild # Switch from Gen 1 to Gen 2 with a clean rebuild ./start-server.sh --gen 2 --rebuild
MCP Tool: search_amplify_docs
The server provides a tool called search_amplify_docs
that can be used to search the Amplify documentation.
Parameters
Parameter | Type | Required | Default | Description |
---|---|---|---|---|
query | string | Yes | - | Elasticsearch query string |
page | number | No | 1 | Page number for pagination |
includeContent | boolean | No | false | Include content snippets in results |
maxResults | number | No | 10 | Maximum number of results to return |
filesOnly | boolean | No | false | Only return file paths without content |
useJson | boolean | No | false | Return results in JSON format |
sessionId | string | No | - | Session ID for related searches |
fullContent | boolean | No | false | Get full content of a specific file |
filePath | string | No | - | Path to a specific file to get full content |
Example Request
{
"query": "authentication react",
"page": 1,
"includeContent": true,
"maxResults": 15
}
Advanced Search Syntax
The search tool supports advanced search syntax:
- Exact phrases:
"authentication flow"
- Exclude terms:
authentication -flutter
- Field-specific search:
title:authentication
- Wildcards:
auth*
- Boolean operators:
authentication AND (react OR javascript) NOT flutter
How the Query Process Works
When you submit a search query, the server processes it through several steps:
-
Query Processing: The server parses your query to understand advanced syntax like boolean operators and wildcards.
-
Directory Optimization: The server uses the directory structure to narrow down which files to search based on your query terms and the selected Amplify generation (Gen 1, Gen 2, or both).
-
Smart Ranking: Results are ranked using a sophisticated algorithm that considers:
- Whether the query mentions specific generations (Gen 1 or Gen 2)
- If the query is about setup, CLI commands, or resource creation
- The relevance of the document to the query context
- Exact matches in document titles
- Match count and document importance
-
Content Extraction: For each matching file, relevant content is extracted and formatted for display.
-
Caching: Search results are cached to improve performance for repeated queries.
This intelligent processing ensures that the most relevant documentation appears at the top of your search results, saving you time and effort.
Generation Selection
The server supports three modes for Amplify documentation generation:
1. Gen 2 Only (Default)
"amplifyGeneration": "gen2"
- Pros: Focused on modern Amplify implementation, reduced disk space usage
- Recommended for: New projects using Amplify Gen 2 features
- Default: This is the default setting as Gen 2 is the current version of Amplify
2. Gen 1 Only
"amplifyGeneration": "gen1"
- Pros: Focused on classic Amplify implementation, reduced disk space usage
- Recommended for: Legacy projects specifically using Amplify Gen 1 features
3. Both Generations
"amplifyGeneration": "both"
- Pros: Complete documentation coverage
- Cons: Larger search area, can cause confusion as Gen 1 and Gen 2 have similar categories
- Use when: You need to reference both generations simultaneously
Project Structure
src/index.ts
: Main server implementationsrc/config.ts
: Configuration loading and processingsrc/git.ts
: Git repository managementsrc/cache.ts
: Search result cachingsrc/directory.ts
: Directory structure managementsrc/types/
: TypeScript type definitionsscripts/build.js
: Build script for preparing documentationbin/mcp
: Executable script for running the server
Recommendations for Usage
-
Optimize for Your Environment:
- For best search results, use
"amplifyGeneration"
as"gen1"
or"gen2"
only
- For best search results, use
-
Search Optimization:
- Use specific technical terms rather than general phrases
- Include category names to narrow results (e.g., "storage owner access" instead of just "access")
- Use quotes for exact phrase matching
- Include abbreviations and alternative terms to improve results
-
Performance Considerations:
- Set an appropriate
autoUpdateInterval
based on your needs (higher values reduce server load) - Use the caching system for frequently accessed queries
- Consider using
filesOnly: true
for initial broad searches to improve performance
- Set an appropriate
Contributing and Feedback
We welcome contributions and feedback to improve this MCP server. If you have suggestions for:
- Improving search query results
- Enhancing the ranking algorithm
- Adding new features or parameters
- Optimizing performance
Please open an issue or submit a pull request on GitHub. Your feedback helps make this tool more effective for everyone. Along the way learn something new.
System Requirements
- Node.js 18.x or higher (tested with Node.js 20.18.2)
- npm 8.x or higher (tested with npm 10.8.2)
License
This project is licensed under the MIT License.