AI Coding Agents

Working with AI Coding Agents

Connect your AI coding agent to EpicContext via the MCP server. Give Claude Code, Cursor, and other agents live access to your product context.

EpicContext connects to AI coding agents through the MCP (Model Context Protocol) server. This gives your agent live, structured access to your product context — brand guidelines, user personas, technical architecture, and more — while it writes code.

The MCP server is the recommended way to work with AI coding agents. It provides a real-time connection, so your agent always works with the latest context.

Prerequisites

Before you begin, make sure you have:

An EpicContext project with content filled in
A CONTEXT folder exported to your project root (via Settings > Export/Import)
Node.js 18 or later installed

The MCP server reads from your local CONTEXT folder and can sync bidirectionally with the EpicContext cloud. Changes made through the MCP are reflected in your web dashboard.

Installation

Install the EpicContext MCP server globally or use it directly with npx:

npx @epiccontext/mcp

No global installation is required. The npx command downloads and runs the latest version automatically.

Connecting to Your AI Agent

Claude Code

Add the MCP server to your Claude Code configuration. In your project's .mcp.json or global Claude config:

{
  "mcpServers": {
    "epicontext": {
      "command": "npx",
      "args": ["@epiccontext/mcp"]
    }
  }
}

Once configured, Claude Code automatically discovers your product context. You can ask it questions like:

"What user personas do we have?"
"What's our tech stack?"
"Read the brand guidelines before styling this component"

Cursor

Add the MCP server to your Cursor settings. Go to Settings > MCP Servers and add:

{
  "mcpServers": {
    "epicontext": {
      "command": "npx",
      "args": ["@epiccontext/mcp"]
    }
  }
}

Cursor will use your product context when generating code, making suggestions that align with your architecture and design decisions.

Other MCP-Compatible Agents

Any AI agent that supports the Model Context Protocol can connect to EpicContext using the same configuration. The MCP server exposes a standard set of tools that work with any compatible client.

Available Tools

Once connected, your AI agent has access to these tools:

Tool	Description
`read_context`	Read all or specific sections of your project context
`get_block`	Retrieve a specific block by its key
`list_blocks`	List all blocks, optionally filtered by section or status
`update_block`	Update a block's content or status
`create_block`	Create a new block in a section
`get_sync_status`	Check sync status between local and cloud
`sync_all`	Synchronize local and cloud content

How It Works

The MCP server runs locally and reads your CONTEXT folder:

Your Code Editor
      ↓ (MCP Protocol)
EpicContext MCP Server
      ↓ (reads/writes)
CONTEXT/ folder in your project
      ↓ (syncs with)
EpicContext Cloud (web dashboard)

This means:

Your AI agent reads context from local files — fast and private
Changes sync to the cloud so your team sees updates in the web dashboard
Edits made in the web dashboard sync back to your local CONTEXT folder

Syncing Content

Pull Latest Changes

To pull the latest context from the cloud to your local folder:

npx @epiccontext/mcp pull

Push Local Changes

To push local changes (including changes made by your AI agent) to the cloud:

npx @epiccontext/mcp push

Automatic Sync

When running through the MCP protocol, the server can sync automatically. Your AI agent can use the sync_all tool to trigger a full bidirectional sync.

Example Workflow

A typical development session with EpicContext and an AI coding agent:

Start your project

Open your code editor. The MCP server starts automatically when your agent connects.

Agent reads context

Your agent reads the relevant context — personas, features, tech stack — before writing code.

Code with context

The agent writes code that follows your brand guidelines, matches your architecture, and serves your user personas.

Agent updates context

As decisions are made during development, the agent can create decision records or update task statuses directly in your context.

Troubleshooting

MCP server not connecting

Make sure your project root contains a CONTEXT/ folder with an AI-GUIDE.md file. If it's missing, export from Settings > Export/Import in the web dashboard.

Context is outdated

Run a manual sync to pull the latest changes:

npx @epiccontext/mcp pull

Agent can't find specific blocks

Check that the block exists in your CONTEXT folder. You can list all blocks with:

npx @epiccontext/mcp list

For the best experience, keep your CONTEXT folder in sync before starting a coding session. A quick npx @epiccontext/mcp pull ensures your agent works with the latest context.

Next Steps

Learn about writing effective context for AI agents
Explore all available block types to enrich your context
Set up the Widget for stakeholder feedback in your application

Last updated: 2026-02-22