MCP 服务器

agentic-store-mcp

A collection of self-hosted, open-source tools that enable AI agents to perform web searching, content crawling, and code analysis tasks like linting and security scanning. It provides power-user capabilities for MCP-compatible clients without requiring external accounts or third-party infrastructure.

README

⚡ Agentic Store MCP: Free, Open-Source Tools for AI Agents

Give your AI agents superpowers with Agentic Store MCP—a collection of free, open-source, and self-hosted tools built on the Model Context Protocol (MCP).

No accounts, no API limits, no subscriptions. Just pure, self-hosted capabilities for your favorite AI assistants.

Fully compatible with: Claude Desktop, Cursor, Windsurf, VS Code, and any other MCP-compatible client.

🚀 Quick Start: Installation

Get started in less than a minute. Choose your preferred environment:

Method 1: Python (Recommended for Developers)

Using uv (the fast Python package installer):

uvx --refresh agentic-store-mcp

Note: The --refresh flag ensures you always run the latest published version.

Need to install uv?

macOS / Linux: curl -LsSf https://astral.sh/uv/install.sh | sh
Windows (PowerShell): powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

Method 2: Docker (No Python Required)

docker run --rm agenticstore/agentic-store-mcp:latest --list

🛠️ MCP Tool Directory

Our tools are organized into powerful modules designed to solve specific workflow challenges for AI agents.

🌐 AgenticData Module

Empower your AI to gather real-time information from the web.

agentic_web_crawl — Advanced Web Scraper & Crawler
- What it does: Fetches any URL and extracts clean page text, SEO metadata, heading structures, links, and images. Handles redirects and encoding automatically.
- Use Case: Let Claude or Cursor read external documentation, scrape competitor sites, or summarize web articles.
- Requirements: None! Works out of the box.
agentic_web_search — Private Metasearch Engine
- What it does: Searches the web via a self-hosted SearXNG instance. Returns ranked results with titles, URLs, and text snippets.
- Use Case: Give your AI live internet access without paying for search APIs.
- Requirements: Requires a local SearXNG instance (docker compose up -d).

💻 AgenticCode Module

Give your AI the ability to analyze, lint, and secure codebases autonomously.

python_lint_checker — Python Static Analysis
- What it does: Checks Python files for imports, bugs, styling issues, and code complexity.
- Use Case: Ask your AI to review your Python code for PEP-8 compliance and logical bugs.
- Requirements: None! Zero external dependencies.
repo_scanner — Security & Secret Scanner
- What it does: Scans directories for leaked secrets (AWS keys, API tokens), PII (emails, SSNs), and .gitignore gaps.
- Use Case: Run a security audit on your project before committing code to public repositories.
- Requirements: None!
dependency_audit — Vulnerability Checker
- What it does: Audits package files (requirements.txt, package.json, go.mod, etc.) for outdated versions and known CVEs using the OSV database.
- Use Case: Ask your AI to check if your project's dependencies are secure and up-to-date.
- Requirements: None!

🔌 Connecting to Your AI Client

Add AgenticStore to your MCP client's configuration file to enable the tools, then restart the client.

<details> <summary><h3>🧠 Claude Desktop</h3></summary>

Config locations:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

Python (All Tools)

{
  "mcpServers": {
    "agentic-store-mcp": {
      "command": "uvx",
      "args": ["--refresh", "agentic-store-mcp"]
    }
  }
}

Docker (All Tools)

{
  "mcpServers": {
    "agentic-store-mcp": {
      "command": "docker",
      "args": ["run", "-i", "--rm", "agenticstore/agentic-store-mcp:latest"]
    }
  }
}

(Check the bottom of this section for SearXNG setup instructions if you want web search enabled!)

</details>

<details> <summary><h3>💻 Cursor</h3></summary>

Config location: ~/.cursor/mcp.json

Python (All Tools)

{
  "mcpServers": {
    "agentic-store-mcp": {
      "command": "uvx",
      "args": ["--refresh", "agentic-store-mcp"]
    }
  }
}

</details>

<details> <summary><h3>🏄 Windsurf</h3></summary>

Config location: ~/.codeium/windsurf/mcp_config.json

(Configuration is identical to the Cursor setup above.)

</details>

<details> <summary><h3>📝 VS Code (MCP Extension)</h3></summary>

(Configuration is identical to the Cursor setup above.)

</details>

🔎 Enabling Web Search (`agentic_web_search`)

To use the web search tool, you need a running SearXNG instance.

Start SearXNG via Docker:

git clone https://github.com/agenticstore/agentic-store-mcp
cd agentic-store-mcp
docker compose up -d

Update your MCP config with the SEARXNG_URL environment variable:

Python users:

{
  "mcpServers": {
    "agentic-store-mcp": {
      "command": "uvx",
      "args": ["--refresh", "agentic-store-mcp"],
      "env": { "SEARXNG_URL": "http://localhost:8080" }
    }
  }
}

Docker users:

{
  "mcpServers": {
    "agentic-store-mcp": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "--network",
        "agentic-store-mcp_default",
        "-e",
        "SEARXNG_URL=http://searxng:8080",
        "agenticstore/agentic-store-mcp:latest"
      ]
    }
  }
}

🎛️ Advanced Usage & CLI Flags

Filter which tools or modules your AI has access to using the CLI:

uvx --refresh agentic-store-mcp                              # Run all tools
uvx --refresh agentic-store-mcp --modules code               # Only load AgenticCode
uvx --refresh agentic-store-mcp --modules data               # Only load AgenticData
uvx --refresh agentic-store-mcp --tools agentic_web_search   # Only load a specific tool
uvx --refresh agentic-store-mcp --list                       # List all available tools and exit

🏗️ Programmatic Usage (Python)

Integrate these MCP tools directly into your own LangChain, LlamaIndex, or custom AI python scripts:

from agentic_store_mcp import start_server

start_server()                              # Load all tools
start_server(modules=["code", "data"])      # Load multiple categories
start_server(tools=["agentic_web_search"])  # Load a single tool

🤝 Contributing & Community

We welcome contributions! To add a new tool:

Ensure the tool is pure Python (no external managed services, completely self-hosted).
Create your tool under agentic_store_mcp/modules/<category>/<tool_name>/ containing:
- handler.py
- schema.json
- README.md
- test_handler.py
Run tests locally using uv run pytest.

📜 License & Support

This project is licensed under the MIT License — free to use, modify, and distribute.

⭐ Enjoying Agentic Store MCP? If you find these tools useful, please star this repository on GitHub to help other developers discover free, open-source MCP tools for their AI workflows!