agentic-store-mcp 0.2.0

Open-source MCP tools for AI agents — lint, security, dependency audit, web scraping and more

agenticstore-mcp-tools

Free, open-source MCP tools for AI agents — self-hosted, no account required.

Connect Claude Desktop, Cursor, Windsurf, VS Code, or any MCP-compatible client. Install everything or only what you need.

---

Modules

Module	Slug	Tools	Status
AgenticCode	code	repo_scanner, dependency_audit, python_lint_checker	Available
AgenticData	data	agentic_web_search	Available
AgenticSearch	search	—	Coming soon
AgenticProductivity	productivity	—	Coming soon
AgenticComms	comms	—	Coming soon
AgenticFinance	finance	—	Coming soon

---

Quick Start

Requirements: Python 3.12+, uv

# 1. Clone
git clone https://github.com/agenticstore/agenticstore-mcp-tools
cd agenticstore-mcp-tools

# 2. Install dependencies
uv sync

# 3. Set up your AI client
uv run setup_local.py

setup_local.py auto-detects Claude Desktop, Cursor, Windsurf, and VS Code — registers the MCP server and injects quality-gate rules into each.

---

Install via PyPI

You can install the tools directly from PyPI.

Using uvx (Recommended)

Run the server without installing it globally. uvx automatically downloads and runs the latest version in an isolated environment.

uvx agentic-store-mcp                      # all tools
uvx agentic-store-mcp --modules code       # AgenticCode only
uvx agentic-store-mcp --modules code,data  # multiple modules

Client config (no cloning needed):

{
  "mcpServers": {
    "agenticstore": {
      "command": "uvx",
      "args": ["agentic-store-mcp", "--modules", "code"]
    }
  }
}

Using pip

You can also install the package globally or in a virtual environment:

pip install agentic-store-mcp

Then configure your AI client to run it:

{
  "mcpServers": {
    "agenticstore": {
      "command": "agentic-store-mcp",
      "args": ["--modules", "code"]
    }
  }
}

Programmatic Usage (Python)

Import and call start_server() directly from your own Python script — no CLI needed.

from agentic_store_mcp import start_server

# Serve all tools
start_server()

# Serve a specific module
start_server(modules=["code"])

# Serve multiple modules
start_server(modules=["code", "data"])

# Serve a single tool
start_server(tools=["agentic_web_search"])

start_server() blocks until the MCP client disconnects (same behaviour as running the CLI). Use it when you want to embed the server in a larger script or manage the process lifecycle yourself.

To use it as a Claude Desktop MCP server via Python:

{
  "mcpServers": {
    "agenticstore": {
      "command": "python3",
      "args": ["-c", "from agentic_store_mcp import start_server; start_server(modules=['code'])"]
    }
  }
}

---

Manual Setup

uv run server.py                        # all tools
uv run server.py --modules code,data    # specific modules
uv run server.py --tools repo_scanner   # specific tool
uv run server.py --list                 # list available tools

---

Connect to Your AI Client

Claude Desktop

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

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

Cursor

~/.cursor/mcp.json

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

Windsurf

~/.codeium/windsurf/mcp_config.json

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

Any MCP-compatible client

Transport: stdio.

uvx agentic-store-mcp --modules code

---

Contributing

1. Deterministic only — pure Python, no LLM calls, predictable output
2. Zero managed dependencies — no AgenticStore infrastructure

---

License

MIT