flowdocs
Transform natural language conversations into professional documentation sites with AI-powered content management
claude mcp add --transport stdio harishsundar-dev-flowdocs node /absolute/path/to/mcp-server/src/simple-server.js \ --env STRAPI_BASE_URL="http://localhost:1337" \ --env STRAPI_API_TOKEN="your_token_here" \ --env DOTENV_CONFIG_SILENT="true"
How to use
FlowDocs is an MCP server that exposes natural-language driven tools to manage and publish documentation. It sits between your Claude Desktop workflow and your Strapi CMS, synchronizing content with a Docusaurus site in real time. The MCP server provides six documentation management tools: create_documentation_page, list_documentation_pages, update_documentation_page, get_documentation_page, delete_documentation_page, and search_documentation. When used through Claude Desktop, you can create and modify articles with natural language and have those changes reflected in Strapi and the Docusaurus site automatically, enabling a seamless content workflow from idea to published docs.
To use it, configure the MCP server in Claude Desktop so Claude can call the tools via JSON-RPC over stdin/stdout. Ensure the Strapi API is accessible (Strapi runs at http://localhost:1337 by default) and that you have a valid API token. Start all services with the provided orchestration script, then prompt Claude with requests like: “Create a documentation page about API authentication” or “List all articles in the AI-generated category.” The system will return structured results (IDs, slugs, content) and perform CRUD operations against Strapi, with content surfaced in the Docusaurus-generated site in near real-time.
If you test end-to-end, you should see Strapi reflect new or updated articles and the Docusaurus site update accordingly as hot-sync runs, ensuring your docs stay coherent across the CMS and the static site frontend.
How to install
Prerequisites
- Node.js v18+ and npm
- Git
- Access to a Strapi instance (default: http://localhost:1337) with an API token
- Clone the repository and navigate to the project root
git clone <repository-url>
cd flowdocs
- Install dependencies for the MCP server
cd mcp-server
npm install
- Configure Strapi and the environment
- Ensure Strapi is running (as described in the readme): http://localhost:1337/admin
- Create or obtain an API token with full access for the Article content type
- Set environment variables for the MCP server as shown in the Claude Desktop setup (STRAPI_BASE_URL, STRAPI_API_TOKEN, DOTENV_CONFIG_SILENT)
- Start all services
cd ..
./start-all.sh
This will launch Strapi, Docusaurus, and the MCP server, exposing the FlowDocs workflow via Claude Desktop integration.
- Configure Claude Desktop
- In Claude Desktop, point the MCP server configuration to the node-based server path and ensure the environment variables above are supplied.
- Restart Claude to load the tools.
- Verify the setup
- Test a simple request to list documents, create a page, and then fetch it by ID or slug to confirm end-to-end operation.
Additional notes
Tips and common issues:
- Ensure the Strapi API token remains valid; regenerate if expired and update the MCP config.
- If you see protocol or 400 errors, verify the payload format for each tool matches the MCP server's expectations (IDs, slugs, and field names).
- The MCP server communicates via JSON-RPC 2.0 over stdio; ensure Claude Desktop and the server process have proper I/O access.
- If hot-sync isn’t updating the Docusaurus site, check that the watcher and syncContent script have access to Strapi and that the Docusaurus build is running.
- Environment variables are critical: STRAPI_BASE_URL must point to your Strapi instance, STRAPI_API_TOKEN must be valid, and DOTENV_CONFIG_SILENT helps avoid noisy dotenv logs in production.
- For production deployments, consider securing tokens and using a dedicated environment management strategy (e.g., .env files or secret managers).
Related MCP Servers
mcp-telegram
MCP Server for Telegram
Remote
A type-safe solution to remote MCP communication, enabling effortless integration for centralized management of Model Context.
mongodb-lens
🍃🔎 MongoDB Lens: Full Featured MCP Server for MongoDB Databases
openapi
OpenAPI definitions, converters and LLM function calling schema composer.
docmole
Dig through any documentation with AI - MCP server for Claude, Cursor, and other AI assistants
the-academy
A Socratic dialogue engine for AI agents.