Up-to-date library docs pulled directly into your AI context.
MCP Servers for Cursor
Cursor ships with native MCP support. Point it at an mcp.json file and your AI editor gains database access, live documentation, browser automation, and more — all via the same protocol Claude Desktop uses.
Set up MCP in Cursor
Four steps to connect your first MCP server.
- 1
Open (or create) your mcp.json
Use ~/.cursor/mcp.json for a global install or .cursor/mcp.json at the root of a project to scope MCPs to that repo.
- 2
Add a server under mcpServers
Each entry needs a command and args. Use the install snippet from any MCP detail page on top-mcps.com.
- 3
Provide credentials via env
For MCPs that need tokens (GitHub, Supabase, Linear), put them in the env block — never inline them in args.
- 4
Restart Cursor
Cmd/Ctrl+Shift+P → "Reload Window". Open the MCP panel to confirm the server is connected.
- Global (all projects)
~/.cursor/mcp.jsonApplies to every workspace you open in Cursor. - Project-scoped
.cursor/mcp.jsonCommit this to share MCPs with your team — checked in at the repo root.
{
"mcpServers": {
"filesystem": {
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-filesystem",
"/Users/you/code"
]
},
"context7": {
"command": "npx",
"args": ["-y", "@upstash/context7-mcp"]
},
"github": {
"command": "npx",
"args": ["-y", "@github/github-mcp-server"],
"env": {
"GITHUB_PERSONAL_ACCESS_TOKEN": "ghp_..."
}
}
}
}Top MCPs for Cursor
The curated starting stack — install these first.
Read and write local files with configurable access controls.
Query and inspect PostgreSQL databases via natural language.
Full browser automation: navigate, click, screenshot, and scrape.
Full Supabase access: database, auth, storage, and edge functions.
All MCPs compatible with Cursor
15 servers verified to work with Cursor, ranked by popularity.
Read and write local files with configurable access controls.
Query and inspect PostgreSQL databases via natural language.
Real-time web search with privacy-focused results.
Persistent knowledge graph memory across AI conversations.
Up-to-date library docs pulled directly into your AI context.
Full Supabase access: database, auth, storage, and edge functions.
Local SQLite database access with full read/write support.
Retrieve web pages and convert them to clean markdown.
Full browser automation: navigate, click, screenshot, and scrape.
Structured step-by-step reasoning for complex problem solving.
Read and send Slack messages, manage channels and threads.
Secure cloud sandboxes for executing AI-generated code.
Manage Linear issues, projects, and cycles from AI context.
FAQ: MCP in Cursor
Where does Cursor store its MCP config?
Cursor reads two files. Global config lives at ~/.cursor/mcp.json. Project-scoped config lives at .cursor/mcp.json in the repo root — useful when you want to share a specific MCP setup with your team via Git.
Do I need to restart Cursor after adding an MCP?
Yes. After editing mcp.json, reload the window (Cmd/Ctrl+Shift+P → Reload Window) or fully quit and reopen Cursor. New servers will not appear until Cursor re-reads the config.
Can I run multiple MCP servers at once in Cursor?
Yes. Every key under mcpServers runs as its own process. A typical Cursor setup combines Filesystem (local files), Context7 (library docs), GitHub (remote repos), and a database MCP — all active simultaneously.
Which MCP is most useful for Cursor users?
Context7. Cursor's built-in model can hallucinate library APIs; Context7 injects real, version-specific documentation before the model answers, which dramatically improves code suggestions. Pair it with Filesystem and GitHub for a solid base.
Does Cursor support MCP servers written in Python?
Yes. MCPs communicate via stdio, so any language works. For Python servers, use "command": "uvx" or "python" in your config instead of "npx".
Browse every MCP server → top-mcps.com