MCP Toolkit
A unified MCP server that provides tools for UI/UX design intelligence, command and agent libraries, and development skills, enabling enhanced design, development, and workflow assistance.
README
MCP Toolkit
Overview
A unified Python MCP (Model Context Protocol) server built on FastMCP that exposes three integrated toolkits to Claude Desktop, Claude Code, and claude.ai: UI/UX Pro Max design intelligence, Claude Command Suite (slash commands + AI agents), and Superpowers development skills. One server, one config, 14 tools spanning design-system generation, stack-specific UI guidelines, browsable command/agent libraries, and composable development skills (TDD, debugging, code review, planning).
Status
Working — server runs via stdio, SSE, or streamable-HTTP; start.sh publishes it through a Cloudflare tunnel for claude.ai integration.
14 tools exposed:
| Group | Tools |
|---|---|
| UI/UX Pro Max | uiux_search, uiux_design_system, uiux_stack_guidelines, uiux_list_domains |
| Command Suite | list_commands, get_command, search_commands, list_agents, get_agent |
| Superpowers | list_skills, get_skill, get_skill_file, search_skills, get_workflow |
Content inventory:
- Commands: 26 namespaces (dev, test, deploy, security, project, rust, svelte, simulation, docs, team, performance, memory, reasoning, wfgy, etc.)
- Agents: 40+ specialized agents (code-auditor, test-engineer, release-manager, svelte-*, swift-macos-expert, wfgy, skill-builder, data-ai, infrastructure, …)
- Skills: 31 Superpowers SKILL.md folders (brainstorming, systematic-debugging, test-driven-development, writing-plans, frontend-design, mcp-builder, webapp-testing, …)
- UI/UX data: CSVs for styles, 161 color palettes, typography pairings, charts, product types, UX guidelines, React/web patterns, Google Fonts, prompt templates — plus
react-nativestack guidelines
Setup & Run
# Install (only FastMCP required)
pip install -r requirements.txt
Claude Desktop (stdio)
fastmcp run server.py
Add to ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"mcp-toolkit": {
"command": "python3",
"args": ["-m", "fastmcp", "run", "/absolute/path/to/mcp-toolkit/server.py"]
}
}
}
Claude Code
claude mcp add mcp-toolkit -- python3 -m fastmcp run /absolute/path/to/server.py
claude.ai (remote via Cloudflare tunnel)
./start.sh # server on :8100 + cloudflared tunnel (prints trycloudflare.com URL)
./start.sh --no-tunnel # local-only
Then in claude.ai: Settings → Integrations → Add Custom MCP → paste the https://…trycloudflare.com/mcp URL.
Manual alternative: fastmcp run server.py -t sse --port 8100 or -t streamable-http.
Architecture
mcp-toolkit/
├── server.py # FastMCP server, 14 @mcp.tool() definitions (UIUX / Commands / Skills)
├── start.sh # Boots streamable-http server + Cloudflare tunnel, prints public URL
├── requirements.txt # fastmcp>=3.0.0
├── agents/ # 40+ agent .md files (flat + subdirs: development, security, business, …)
├── commands/ # 26 namespaces of slash-command .md files (dev, test, deploy, rust, svelte, …)
├── skills/ # 31 Superpowers skills, each a folder with SKILL.md + supporting files
├── uiux_data/ # CSVs: styles, colors, typography, charts, products, UX, stacks
└── uiux_scripts/ # core.py, design_system.py, search.py — imported by server.py
Wiring note: server.py patches sys.path and core.DATA_DIR so the bundled uiux_scripts/ modules resolve uiux_data/ regardless of cwd.
Tool surface:
uiux_search(query, domain, max_results)— 12 domains auto-detecteduiux_design_system(query, project_name)— full design-system markdown (colors/typography/spacing/components)uiux_stack_guidelines(query, stack)— currentlyreact-nativelist_commands/get_command/search_commands— walkcommands/<ns>/*.mdlist_agents/get_agent— walkagents/**/*.md, parse frontmatterdescription:list_skills/get_skill/get_skill_file/search_skills— walkskills/*/SKILL.mdget_workflow(workflow)— curated multi-skill recipes forfeature | bugfix | refactor | review
Roadmap
- [ ] Expand
uiux_stack_guidelinesbeyondreact-native(React, Next.js, Vue, Svelte, SwiftUI, Flutter, Tailwind, shadcn/ui, HTML/CSS) - [ ] Cache
_get_agents()/_get_namespaces()/SUPERPOWERSwalks — currently re-scan disk on every tool call - [ ] Make
SUPERPOWERSset construction lazy; it runs at import time and will crash ifskills/is ever missing - [ ] Replace hardcoded Python path in
start.sh(/Library/Frameworks/Python.framework/Versions/3.11/bin/python3) with a venv or$(command -v python3) - [ ] Auth / access control for the public Cloudflare tunnel endpoint
- [ ] Health check + reconnect logic in
start.sh(currently fixedsleep 8for tunnel URL parse) - [ ] Add MCP resources (not just tools) for large static content — design-system CSVs, agent/skill catalogs
- [ ] Tests — no test suite exists yet
Known Bugs
- None identified in shipped tool code. (TODO markers found in
commands/skills/scripts/init_skill.pyandquick_validate.pyare intentional template scaffolding for the skill-builder, not defects.) - Minor:
SUPERPOWERS = {…}at module top-level will raiseFileNotFoundErrorifskills/is absent — deployment-time concern only. - Minor:
start.shparses the tunnel URL after a fixed 8-second sleep; slow cloudflared startups may leaveTUNNEL_URLempty.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.