My

You learn to run and use multiple MCP servers that extend Cursor with Git, Brave Search, GitHub, Memory storage, and Sequential Thinking capabilities. These servers connect via standard MCP interfaces to provide project-wide repository actions, rich search, persistent memory, and structured problem solving, all accessible through your MCP client.

How to use

Use your MCP client to connect to the available MCP servers using the stdio configurations described below. Each server exposes a dedicated, named entry you can enable in Cursor Settings > Features > MCP. Once connected, you can delegate tasks such as repository operations, web and local searches, memory-backed retrieval, and step-by-step thinking to the respective MCPs. You can mix and match several servers to compose complex workflows: for example, perform a Git operation, run a Brave Search query, and then reason through a problem with sequential thinking all in a single session. Ensure your API keys and tokens are correctly set in your environment so the servers can access their external services.

How to install

Prerequisites must be met before installation. You need an appropriate operating system, Python 3.10 or higher, Node.js 18 or higher, and write permissions in the installation directory.

# Phase 1: system setup
make phase1

Restart your terminal, then proceed to install MCP components in phase 2.

# Phase 2: MCP installation
make phase2

Optional maintenance commands you may use during or after installation.

make update-all    # Update all MCP servers
make clean         # Remove installed dependencies
make clean-deep    # Complete system cleanup

Configuration and usage details

Set up required API keys before starting the servers. You will need a Brave Search API key and a GitHub Personal Access Token. Store them in your environment so the MCPs can authenticate to external services.

export BRAVE_API_KEY=your_key_here

export GITHUB_PERSONAL_ACCESS_TOKEN=your_token_here

MCP connection configurations

Configure each MCP in Cursor Settings to define how it runs and connects. Below are the explicit stdio configurations shown for each MCP server. Copy these into Cursor as-is to enable the servers.

{
  "name": "Git",
  "type": "stdio",
  "command": "uvx mcp-server-git --repository /path/to/your/repo"
}

{
  "name": "Brave Search",
  "type": "stdio",
  "command": "npx -y @modelcontextprotocol/server-brave-search"
}

{
  "name": "GitHub",
  "type": "stdio",
  "command": "npx -y @modelcontextprotocol/server-github"
}

{
  "name": "Memory",
  "type": "stdio",
  "command": "npx -y @modelcontextprotocol/server-memory"
}

{
  "name": "Sequential Thinking",
  "type": "stdio",
  "command": "npx -y @modelcontextprotocol/server-sequential-thinking"
}

Available tools

git_repo_operations

Interact with a local Git repository: view status, manage files, and perform repository operations.

git_branch_and_commit

Manage branches and perform commit operations, including creating, switching, and merging branches.

git_diff

Compute diffs for staged and unstaged changes, or compare differences between branches.

brave_search_web_local

Conduct web searches with Brave Search and perform local business, news, and article searches.

github_repo_management

Automate GitHub repositories: manage repos, issues, and pull requests with MCP integration.

memory_graph_management

Store, retrieve, and query a knowledge graph with entities and relations for memory-backed workflows.

sequential_thinking

Structure problem solving with step-by-step analysis, dynamic thought revision, and context maintenance.