Meta

• Author: K-Dense AI
• Maintainer: K-Dense AI

Meta

• License: Apache Software License (Apache-2.0)
• Tags agent , ai , claude , mcp , proxy , skills
• Requires: Python <3.13, >=3.12
• Provides-Extra: test

Classifiers

• Development Status
  • 4 - Beta
• Intended Audience
  • Developers
  • Science/Research
• License
  • OSI Approved :: Apache Software License
• Programming Language
  • Python :: 3.12
• Topic
  • Scientific/Engineering :: Artificial Intelligence

Claude Skills MCP Frontend

Lightweight MCP proxy for Claude Skills that auto-downloads the heavy backend on demand.

Overview

This is the frontend component of the Claude Skills MCP system. It's a lightweight proxy (~15 MB) that:

• Starts instantly (<5 seconds)
• Auto-downloads the backend when first needed
• Acts as MCP server (stdio) for Cursor
• Acts as MCP client (HTTP) for the backend
• Returns tool schemas immediately (no backend wait needed)

Installation

# Via uvx (recommended for Cursor)
uvxclaude-skills-mcp

# Via uv tool (persistent install)
uvtoolinstallclaude-skills-mcp

# Via pip
pipinstallclaude-skills-mcp

Usage with Cursor

Add to your Cursor MCP settings (~/.cursor/mcp.json):

{
"mcpServers":{
"claude-skills":{
"command":"uvx",
"args":["claude-skills-mcp"]
}
}
}

Restart Cursor and the skills will be available!

First Run Behavior

On first run, the frontend will:

1. Start immediately (~5 seconds) ✅ Cursor timeout satisfied!
2. Return tool schemas to Cursor (instant)
3. Download backend in background (~250 MB, 60-120 seconds)
4. When you first use a tool, you'll see "Loading backend..."
5. Once backend ready, all tools work normally

Subsequent runs: Fast! Backend is already installed.

Configuration

The frontend forwards all arguments to the backend:

# Custom configuration
uvxclaude-skills-mcp--configmy-config.json

# Verbose logging
uvxclaude-skills-mcp--verbose

# Custom backend port (advanced)
uvxclaude-skills-mcp--port9000

Remote Backend (Future)

# Connect to hosted backend instead of local
uvxclaude-skills-mcp--remotehttps://skills.k-dense.ai/mcp

Note: Remote backend support coming in v1.1.0

How It Works

Cursor → Frontend (stdio, ~15 MB)
 ↓
 list_tools() → Returns hardcoded schemas INSTANTLY ✅
 ↓
 [Backend downloads in background...]
 ↓
 call_tool() → Proxies to Backend (HTTP)
 ↓
 Backend (HTTP, ~250 MB) → Performs actual search

This architecture solves the Cursor timeout problem by separating:

• Fast startup (frontend, minimal dependencies)
• Heavy processing (backend, downloads async)

Dependencies

Frontend only requires:

• mcp>=1.0.0 (~5 MB)
• httpx>=0.24.0 (~5 MB)

Total: ~15 MB (downloads in <10 seconds)

The backend (claude-skills-mcp-backend) is auto-installed on first use.

Troubleshooting

"Backend not ready" message

On first run, you'll see this message for 30-120 seconds while the backend downloads. This is normal and only happens once.

Backend installation fails

Check:

1. Internet connection
2. Disk space (~500 MB free needed)
3. Python 3.12 installed

Tools not working

Run with verbose logging:

uvxclaude-skills-mcp--verbose

Check logs in stderr for backend status.

Development

# Clone the monorepo
gitclonehttps://github.com/K-Dense-AI/claude-skills-mcp.git
cdclaude-skills-mcp/packages/frontend

# Install in development mode
uvpipinstall-e".[test]"

# Run tests
uvrunpytesttests/

Related Packages

• claude-skills-mcp-backend (Backend): Heavy server with vector search
• Main Repository: https://github.com/K-Dense-AI/claude-skills-mcp

License

Apache License 2.0

Copyright 2025 K-Dense AI (https://k-dense.ai)

Verified details

These details have been verified by PyPI

Maintainers

👁 Avatar for OrionLi545 from gravatar.com
OrionLi545 👁 Avatar for timothykassis-k-dense from gravatar.com
timothykassis-k-dense

Unverified details

These details have not been verified by PyPI

Project links

• Backend
• Bug Tracker
• Documentation
• Homepage
• Repository

• Download URL: claude_skills_mcp-1.0.6.tar.gz
• Upload date: Oct 24, 2025
• Size: 23.1 kB
• Tags: Source
• Uploaded using Trusted Publishing? No
• Uploaded via: uv/0.8.12

Hashes for claude_skills_mcp-1.0.6.tar.gz

Algorithm	Hash digest
SHA256	80bb2bdfb03c4fed69878ae3dfea20beeb10ae9b3e633a3cd69c18facc063884
MD5	b98998cd86f04b6fad0d6b37b7edef1a
BLAKE2b-256	7efb48f91721f8371bc999f672c893d0fa979944f340eba6d42bbef967e7c2c1

See more details on using hashes here.

• Download URL: claude_skills_mcp-1.0.6-py3-none-any.whl
• Upload date: Oct 24, 2025
• Size: 12.6 kB
• Tags: Python 3
• Uploaded using Trusted Publishing? No
• Uploaded via: uv/0.8.12

Hashes for claude_skills_mcp-1.0.6-py3-none-any.whl

Algorithm	Hash digest
SHA256	2846d349391af4094d59fb97cd1c2ec860ed986e73c70c2724f1a38bd1922c25
MD5	a200a0e39f0eea2ff049efa202a99a18
BLAKE2b-256	24cfedb94a574d79fafed971190abe3400ad22024248faff103730909d063b55

See more details on using hashes here.