Meta

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

Meta

• License: Apache Software License (Apache-2.0)
• Tags agent , ai , backend , claude , mcp , scientific , skills , vector-search
• 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 Backend

Heavy backend server for Claude Skills MCP system with vector search capabilities.

Overview

This is the backend component of the Claude Skills MCP system. It provides:

• Vector-based semantic search using sentence-transformers
• Skill indexing and retrieval
• MCP protocol via Streamable HTTP transport
• Background skill loading from GitHub and local sources

Note: This package is typically auto-installed by the frontend (claude-skills-mcp). You only need to install it manually for:

• Remote deployment (hosting your own backend)
• Development and testing
• Standalone usage without the frontend proxy

Installation

# Via uv (recommended)
uvtoolinstallclaude-skills-mcp-backend

# Via uvx (one-time use)
uvxclaude-skills-mcp-backend

# Via pip
pipinstallclaude-skills-mcp-backend

Usage

Run Standalone Server

# Default (localhost:8765)
claude-skills-mcp-backend

# Custom port
claude-skills-mcp-backend--port8080

# For remote access
claude-skills-mcp-backend--host0.0.0.0--port8080

# With custom configuration
claude-skills-mcp-backend--configmy-config.json

# Verbose logging
claude-skills-mcp-backend--verbose

Configuration

# Print example configuration
claude-skills-mcp-backend--example-config>config.json

# Edit config.json to customize skill sources, embedding model, etc.

# Run with custom config
claude-skills-mcp-backend--configconfig.json

Endpoints

When running, the backend exposes:

• Streamable HTTP MCP: http://localhost:8765/mcp
• Health Check: http://localhost:8765/health

Docker Deployment

Build Image

dockerbuild-tclaude-skills-mcp-backend.

Run Container

# For local access
dockerrun-p8765:8765claude-skills-mcp-backend

# For remote access
dockerrun-p8080:8765\
-eHOST=0.0.0.0\
claude-skills-mcp-backend--host0.0.0.0--port8765

Dependencies

This package includes heavy dependencies (~250 MB):

• PyTorch (CPU-only on Linux): ~150-200 MB
• sentence-transformers: ~50 MB
• numpy, httpx, fastapi, uvicorn: ~30 MB

First download may take 60-180 seconds depending on your internet connection.

Performance

• Startup time: 2-5 seconds (with cached dependencies)
• First search: +2-5 seconds (embedding model download, one-time)
• Query time: <1 second after models loaded
• Memory usage: ~500 MB

Development

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

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

# Run tests
uvrunpytesttests/

Related Packages

• claude-skills-mcp (Frontend): Lightweight proxy that auto-installs this backend
• 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

• Bug Tracker
• Documentation
• Homepage
• Repository

• Download URL: claude_skills_mcp_backend-1.0.6.tar.gz
• Upload date: Oct 24, 2025
• Size: 78.3 kB
• Tags: Source
• Uploaded using Trusted Publishing? No
• Uploaded via: uv/0.8.12

Hashes for claude_skills_mcp_backend-1.0.6.tar.gz

Algorithm	Hash digest
SHA256	6954ea824a15182c485265e32709ebccce73516d0f3353d5fc983cc8952cc0f4
MD5	b3d4825b5a0aa2097b563e7c16e896df
BLAKE2b-256	5e9127acabbf6c890627caf28164ac230f9de57c8a73f8e99dacab3c90779e60

See more details on using hashes here.

• Download URL: claude_skills_mcp_backend-1.0.6-py3-none-any.whl
• Upload date: Oct 24, 2025
• Size: 31.4 kB
• Tags: Python 3
• Uploaded using Trusted Publishing? No
• Uploaded via: uv/0.8.12

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

Algorithm	Hash digest
SHA256	7c522651c5bbffbdecd1200ad08fb009cf45cfc5f23281a99cafd78ca1362afa
MD5	331fecbc0417a4e51abd7a0512945ef9
BLAKE2b-256	3ef81fbac56a2330bc35bc5b288f67b1ae8586565a9d5e78dc9cdfb1abbbfcde

See more details on using hashes here.