Amplify Docs Simple

A simple MCP server that serves pre-filtered AWS Amplify Next.js documentation. No complex transformations, no JSON generation, just clean MDX content.

What's Included

• 883 AWS Amplify documentation files
• Pre-filtered for Next.js content only
• Full-text search with 10,000+ indexed keywords
• Searches actual content, not just file names
• Removes iOS/Android/Flutter specific code
• Keeps universal documentation
• 6MB total size (down from 17MB)

Philosophy

Instead of complex MDX parsing and transformation:

1. Pre-filter AWS docs for Next.js content
2. Build a simple search index
3. Serve the MDX files directly
4. Let AI understand the natural documentation

Installation

# Install globally
npm install -g @chinchillaenterprises/amplify-docs-simple

# Or use with npx
npx @chinchillaenterprises/amplify-docs-simple

Usage with Claude

# Add to Claude
claude mcp add amplify-docs-simple -s user -- npx @chinchillaenterprises/amplify-docs-simple

Tools

Only 3 simple tools:

search_amplify_docs

Full-text search across all documentation content.

• query (required): Search terms (e.g., "schedule lambda function", "EventBridge", "cron")
• category (optional): Filter by auth, storage, data, functions, ai, general

Now searches inside documentation content, not just titles!

get_amplify_doc

Get the full MDX content of a documentation file.

• path (required): Path from search results

list_amplify_categories

List all categories and document counts.

Example

Human: How do I implement social login?

Claude: Let me search for social login documentation.
[Calls search_amplify_docs with query "social login auth"]

Here are the relevant docs:
1. **Add social provider sign-in**
   Category: auth
   Path: `src/pages/[platform]/build-a-backend/auth/add-social-provider/index.mdx`

[Calls get_amplify_doc with that path]

[Shows the full MDX content with all examples, prerequisites, and context]

Why This Works Better

1. No Information Loss - All context, examples, and relationships preserved
2. Simpler - ~200 lines of code vs 2000+
3. Faster - No complex parsing/transformation
4. More Accurate - AI sees docs exactly as AWS wrote them
5. Easier Updates - Just pull new MDX files

License

MIT