Practical overview
When it fits
Use Ragrabbit when you want to turn your own website or documentation into a searchable AI knowledge source and can provide the required database and OpenAI configuration.
Who it is for
- Website and documentation teams building AI-assisted site search.
- Operators who want to self-host a searchable index of their website content.
- Teams that want supported AI clients to retrieve documentation through MCP.
Documented uses
- Index one URL or recursively crawl a website.
- Offer visitors an embedded AI chat or search interface over indexed pages.
- Generate and preview a customizable LLM.txt file with a reorderable table of contents.
- Let supported AI clients retrieve documentation pages using semantic search.
- Replace Fumadocs search with Ragrabbit components or add its floating chat button.
Requirements & limits
The project is open source under the MIT license and supports self-hosting. Its documented requirements include Node.js 20.x, PostgreSQL with pgVector, and an OpenAI API key. The repository provides a Vercel deployment flow.
- The indexing workflow requires starting the Job Runner and keeping its tab open until the job finishes.
- The MCP server is described for supported AI clients, with Claude Desktop and Cursor IDE named in the README.
- The documented Fumadocs integration uses the @ragrabbit/search-react package.
- The Vercel setup flow calls for PostgreSQL integration and admin authentication configuration.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Index a website and preview its LLM.txt
Configure Ragrabbit to crawl a website and inspect the generated LLM.txt.
Before you start
- Node.js 20.x
- PostgreSQL with pgVector
- OpenAI API key
- A deployed or locally running Ragrabbit instance with admin access
Steps
Set the OpenAI API key and configure admin authentication.
In the Indexing section, add a single URL or choose a website for recursive crawling.
Start the Job Runner and keep its tab open until the job finishes.
Open the LLM.txt section to preview the generated file.
Expected outcome
The selected pages are indexed and the generated LLM.txt can be previewed in the application.
Guide limitations
- The Job Runner tab must remain open until indexing finishes.
- The guide covers indexing and previewing; it does not document publishing the generated file to a site.
Embed Ragrabbit search in a website
Add the documented search widget to a site after Ragrabbit is deployed and content is indexed.
Before you start
- A deployed Ragrabbit instance
- Website content indexed in Ragrabbit
- Ability to edit the site's HTML or React application
Steps
Add the Ragrabbit widget script to the page.
For the documented search-input embed, include the search script and the element.
For a React project, use the documented Next.js Script component and mount the custom element.
Expected outcome
The page includes a Ragrabbit search input connected to the indexed content.
Guide limitations
- The README’s plain HTML example uses the ragrabbit.com script host; it also documents a deployed-app URL pattern for the chat button.
- The React example is specifically written for a Next.js client component.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 12 October 2026
- Original website title
- madarco/ragrabbit · Official homepage title · retained in its source language