The Neobrutalism MCP serverin JetBrains IDEs
JetBrains AI Assistant’s MCP settings take an HTTP server JSON block. Add Neobrutalism once and the same tools show up in AI chat across IntelliJ, WebStorm, PyCharm, and the rest of the family — catalog search, real block source for your backend, and install commands with your registry token already in the URL. No local MCP process to babysit.
neobrutalism · JetBrains
Add a hero to the homepage
JetBrainsviaMCP
Used Neobrutalism MCP to find a hero that fits the product.
Run your team
in one place
Plans, docs, and deadlines — shared with everyone who needs them.
Get StartedGet All Access
1 promptHero shipped
- MCP server
- Built for AI agents
- 12 tools
- 150+ Pro blocks
- 50+ components
- 3 backends
- Claude Code
- Cursor
- VS Code
- Any MCP client
- One URL, one token
- Prompt → shipped
One command
to connect
Add the server to your editor
Settings → Tools → AI Assistant → MCP
JetBrains setup
FAQ
Connect JetBrains AI Assistant to Neobrutalism MCP — HTTP MCP settings, IntelliJ/WebStorm/PyCharm, token, and Pro.
Go to Settings → Tools → AI Assistant → Model Context Protocol (MCP) → +. Choose the HTTP / Streamable HTTP connection, paste a JSON snippet with `neobrutalism` and your registry `url` (token in the query string), set server level (Global or Project), then Apply. Status should show connected; open the tools icon to confirm 12 tools.
MCP config lives in AI Assistant settings per IDE. Add the same URL/token in each IDE you use, or use Settings Sync so the config travels. The Neobrutalism endpoint itself is the same everywhere.
In AI Assistant chat, ask: “what neobrutalism block categories exist?” The assistant should call list_block_categories. You can also type `/` in chat to see MCP commands from the server once tools are loaded.
Mention the server by name (“use the neobrutalism MCP tools”) or invoke a tool via `/`. Check MCP status in settings: 401 bad token, 403 no Pro. Enable automatic start for new/changed servers if the entry never launched after Apply.
Yes — AI Assistant can Import from Claude on the MCP settings page. After import, still point the Neobrutalism entry at your Pro registry URL with your own token; don’t reuse someone else’s secret.
Yes — AI Assistant only gets Pro registry tools when the Neobrutalism account is Pro. No token → 401; token without Pro → 403. Upgrade and reconnect the HTTP server in MCP settings; same token works.
Dashboard → Account. Same personal registry token as Pro shadcn CLI installs.
Global is best for a personal token used in every project. Use Project only if you want the server limited to one codebase — still don’t commit the token; keep secrets out of VCS.
Team/Org Pro seats each have their own token. Every developer adds the HTTP server in their IDE with their own secret (or via Settings Sync of a non-secret template plus a local token).
Search the catalog, open real block/component source, fetch previews, and generate token-ready `shadcn add` commands while you stay in IntelliJ, WebStorm, PyCharm, or another JetBrains IDE with AI Assistant.
The bundled IDE MCP Server exposes *your IDE* to external agents (Claude Code, Codex, etc.). Neobrutalism MCP is the opposite direction: AI Assistant consumes *our* hosted registry. You can run both; they solve different problems.
It can retrieve install commands via get_install_command. Run the returned `shadcn add` line in the IDE terminal to install the block and dependencies for your backend.
Help → Show Log in Explorer/Finder, then open the `mcp` folder for per-server logs. Status column errors in MCP settings usually map to 401 (token) or 403 (Pro).
Ask for the backend you use (default Base UI). Source is flattened before return so it matches a normal shadcn install.