AG Grid’s new Model Context Protocol (MCP) Server - AG MCP - provides LLMs with version and framework-specific AG Grid context. This ensures responses use the most accurate and up-to-date information.
AG MCP is open source and works with any LLM that supports MCP, such as Claude, Cursor, and Copilot:
What is Model Context Protocol (MCP)?
Model Context Protocol (MCP) is an open-source protocol, proposed by Anthropic, for connecting AI applications to external systems. At its core, MCP is about providing additional, highly relevant context to LLMs so that they can more accurately and efficiently assist developers with specific tasks.
Using MCP, AI applications like Claude or ChatGPT can connect to data sources (e.g. docs), tools (e.g. search engines) and workflows (e.g. specialised prompts), enabling them to access key information and perform tasks, or, as Anthropic put it:
Think of MCP like a USB-C port for AI applications. Just as USB-C provides a standardised way to connect electronic devices, MCP provides a standardised way to connect AI applications to external systems.
The Problem: LLMs Require the Latest and Most Accurate Data Possible
As LLM-powered assistants become more integrated into developer workflows, their usefulness depends on the freshness and accuracy of their knowledge.
AG Grid has over 350,000 documentation pages across 99 versions and 4 frameworks, with new content added every six weeks for minor releases and twice a year for major ones.
This constant evolution creates several challenges for LLMs:
- Outdated training data – missing recent features, APIs, or configuration changes.
- Framework nuances – React, Angular, Vue, and Vanilla integrations each require a different setup and use a different syntax.
- Version-specific differences – migration guides, performance tips, and configuration options vary between releases.
- Residual context – information removed from the docs may still persist in an LLM’s training data.
The result is that even advanced assistants can return outdated or incorrect AG Grid guidance.
The Solution: AG MCP
The AG MCP Server addresses this by integrating directly with your LLM client, providing it with access to live, authoritative AG Grid knowledge.
At its core, AG MCP provides an LLM-optimised search interface that delivers version and framework-specific documentation, examples, and API references in condensed Markdown. This ensures responses remain precise while minimising token usage.
The AG MCP Server is composed of three key components:
- Tools - Schema-defined interfaces that enable AI models to perform actions, including searching the docs and detecting & setting the current AG Grid version.
- Prompts - Pre-configured actions that allow you to perform common actions, such as creating new grids or migrating to later versions.
- Resources - Documentation, examples, and API definitions/interfaces in condensed markdown.
Each of these components are available to your LLM whenever it needs more information on how to implement AG Grid features, and can also be accessed directly.
Example Usage
At a high level, the AG MCP Server works as follows:
- Prompt is submitted to LLM client, e.g. Claude, Cursor, Copilot, ChatGPT, etc...
- LLM accesses
ag-mcpserver and requests additional context, via a tool call, pre-built prompt, or by requesting specific resources. ag-mcpuses an LLM optimised search tool to return version-specific information, such as docs, examples, and api refs in condensed markdown.- LLM uses this additional context to provide more accurate responses.
Example Use Cases
Whilst the AG MCP will ensure your LLM accesses the latest, most accurate information for any use case, some we think will be particularly interesting include:
- Faster Initial Configuration
Quickly scaffold a new grid tailored to your framework and version. Instead of piecing together examples, you can use thequick-startprompt when creating a new AG Grid project or adding AG Grid to an existing project. - Simplified Version Migration
Automatically retrieve migration guidance when upgrading between AG Grid versions. Theupgrade-gridprompt creates a step by step plan to help migrate from your current version to the provided version. This is given to the LLM to execute, calling back to the MCP server as needed. It takes a version by version approach, making sure each version is correct before continuing. - Specific Feature Implementation
Get precise instructions and examples for building custom themes with the Theming API, or implement complex features, such as Pivoting or the Server Side Row Model, into existing instances of AG grid.
Getting Started
To install AG MCP in your LLM, use the npx ag-mcp command during the MCP install process, as defined by your LLM; for example, to add ag-mcp to Claude Code, run the following command:
claude mcp add ag-mcp npx ag-mcpSummary
The AG MCP will reduce the amount of time needed to integrate, configure, and maintain AG Grid so that you can focus on your other tasks, but we need your help to make it even better.
If you'd like to share any feedback, enterprise customers can get in touch with us via Zendesk, and our Community users can raise an issue on the ag-mcp repo, which is monitored by our product team.
You can start using AG MCP today. Install it with npx ag-mcp and follow our docs to connect it to your LLM: