agentic-store-mcp 0.6.0

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

⚡ AgenticStore MCP: The Ultimate Open-Source AI Agent Toolkit

Supercharge your AI Assistant with 27 Powerful Model Context Protocol (MCP) Tools

🚀 Quick Start • 📋 Full Tool Directory • 🌐 Web Search • 🔌 Client Setup • ⭐ Star Repo

---

AgenticStore MCP turns Claude Desktop, Cursor, and Windsurf into unstoppable autonomous agents. Get instant access to web search, codebase analysis, GitHub integration, and persistent memory—all locally hosted, with no subscriptions or vendor lock-in.

🔥 Why Choose AgenticStore MCP?

• 🔒 100% Privacy-First: Everything runs locally. Your code and data never leave your machine.
• 💸 Truly Free: No accounts, no paywalls, no subscriptions.
• 🧠 Persistent Agent Memory: Let your AI remember facts and contexts across sessions.
• ⚡ Plug & Play: Installs in seconds via uvx or Docker.

---

📋 Table of Contents

• 🧰 What's Inside the Toolkit
• 🚀 Quick Start Guide (3 Ways to Install)
  • V0: Python/uvx (Fastest)
  • V1: Docker (Isolated)
  • V2: MCP Hub Webapp (UI)
• 🔌 Connect to Your AI Client
• 🗂 Tool Directory — All 27 Tools
• 🔎 Enabling Web Search
• 🎛 Advanced Usage & CLI Flags
• 🤝 Contributing

---

🧰 What's Inside the Toolkit

Equip your AI client with these 4 powerhouse modules containing 27 specific tools:

Module	Purpose	Key Capabilities	Tools
💻 Code & Security	Codebase mastery & safety	Static analysis, CVE scans, Git commit summaries	11
🌐 Data & Search	Web access	Private web search (SearXNG) and deep web crawling	2
🧠 Memory	Agent context persistence	Save/read facts, session checkpoints, changelogs	12
🔧 Meta-Tools	Configuration & discovery	Tool discovery, runtime config management	2

---

🚀 Quick Start Guide (3 Ways to Install)

Pick the setup that best fits your workflow. Don't know which to pick? Start with V0.

V0 — Python / uvx (Fastest start, no Docker needed)

1️⃣ Install uv (one-time setup)

# 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"

2️⃣ Run the MCP server

uvx --refresh agentic-store-mcp

(The --refresh flag ensures you always pull the latest published stable version).

3️⃣ Configure your AI Client
See the Connect to Your AI Client section to link it!

💡 Pro-Tip (Web Search): Want web search? Add "env": { "SEARXNG_URL": "http://localhost:8080" } to your config and run docker compose up -d searxng from this repo. See Web Search Setup.

V1 — Docker (No Python required, fully isolated)

1️⃣ Ensure Docker Desktop is running

2️⃣ Clone and start services

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

This starts a private web search engine (SearXNG) at http://localhost:8080 and the AgenticStore Webapp at http://localhost:8765.

3️⃣ Client configuration
Open http://localhost:8765 → Clients tab to get your exact copy-paste config snippet. Example for Claude:

{
  "mcpServers": {
    "agentic-store-mcp": {
      "command": "docker",
      "args": [
        "run", "-i", "--rm",
        "-v", "agentic-store-mcp_agentic-store-memory:/root/.config/agentic-store",
        "--network", "agentic-store-mcp_default",
        "-e", "SEARXNG_URL=http://searxng:8080",
        "agentic-store-mcp"
      ]
    }
  }
}

V2 — MCP Hub UI (Manage everything visually)

Forget manual JSON editing! Use our local web UI to:

• 🔑 Connectors: Enter GitHub/OpenAI API keys securely (stored in OS keyring).
• 🛠️ Tools: Toggle which of the 27 tools to expose to the AI.
• 💻 Clients: Auto-generate configuration for Claude, Cursor, and Windsurf.
• 🧠 Memory: Manage persistent agent states, checkpoints, and logs.

Start the Hub via Python:

git clone https://github.com/agenticstore/agentic-store-mcp
cd agentic-store-mcp
uv sync
uv run webapp.py

(Or via Docker: docker compose up -d)

Access it at: http://localhost:8765

---

🔌 Connect to Your AI Client

Add the configuration snippet to your respective client's config file. Remember to restart the client after saving!

Client	Config File Path
Claude Desktop (Mac)	~/Library/Application Support/Claude/claude_desktop_config.json
Claude Desktop (Win)	%APPDATA%\Claude\claude_desktop_config.json
Cursor	~/.cursor/mcp.json
Windsurf	~/.codeium/windsurf/mcp_config.json
VS Code	Appends to your VS Code settings.json under MCP extension config

The Config Snippet (for V0 uvx setup):

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

⬆️ Back to Top

---

🗂 Tool Directory — All 27 Tools

💻 Code Tools & Integrations (11 tools)

Codebase Analysis

Analyze, search, and navigate your codebase flawlessly.

Tool	Capability
python_lint_checker	Runs static analysis on Python files (finds bugs, unused imports, complexity). No external deps.
search_code	Blazing-fast full-text search across local codebases with regex & file-type filtering.
get_file	Read files from defined paths, including specific line-range slicing.
analyze_commits	Contextualize agents with recent repo history (diff stats, authors, messages).

GitHub Integration

(Requires a GitHub Personal Access Token. Set via GITHUB_TOKEN or MCP Hub).

Tool	Capability
get_repo_info	Fetch metadata: stars, forks, primary language, open issues.
manage_issue	Create, comment on, close, or list issues in accessible repositories.
create_pr	Automatically open new internal Pull Requests with title & body definitions.

Security & Auditing

Agent-driven DevSecOps.

Tool	Capability
repo_scanner	Detects leaked secrets (API keys), PII, and validates .gitignore.
dependency_audit	Scans requirements.txt, package.json, etc. against the OSV CVE database.
code_scanning_alerts	Fetches active CodeQL/Security alerts from GitHub.
dependabot_alerts	Fetches Dependabot vulnerability alerts from GitHub.

🌐 Data & Search (2 tools)

Tool	Capability
agentic_web_crawl	Extract clean markdown text, headings, and metadata from any URL. Handles redirects beautifully.
agentic_web_search	Conduct live web searches via self-hosted SearXNG. Returns structured snippets.

🧠 Memory & Agent Orchestration (12 tools)

Persistent memory lets AI agents hand off work across sessions and restarts. 100% locally stored.

Storage Primitives

Tool	Capability
memory_write	Store a key-value fact in local JSON. Survives restarts.
memory_read	Retrieve specific (or all) stored facts.
memory_search	Fuzzy full-text search across all stored facts.
memory_log	Append timestamped entries to an immutable session JSONL log (decision trails).
memory_checkpoint	Save a complete snapshot: state, decisions, next steps, and client context.
memory_restore	Load a named checkpoint back into the active agent context.

Productivity Layer

Tool	Capability
spinup_memory	Initialize a new project context (stub files for plans, milestones, learnings).
restore_session	One-call holistic session restore (loads latest checkpoint, plans, logs, and facts).
update_plan	Create or update the active plan.md for task tracking.
update_milestones	Append or patch individual milestone progress via regex.
update_learnings	Log technical discoveries into a growing context directory.
update_change_log	Append semantic release notes to CHANGELOG.md automatically.

🔧 Toolkit Meta-Tools (2 tools)

Tool	Capability
tool_search	List all available toolkit tools, requirements, and descriptions.
configure	Dynamically override runtime configurations and module states.

---

🔎 Enabling Web Search

To give your agent internet access, agentic_web_search uses a private SearXNG instance.

1. Start the SearXNG Docker container:

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

2. Pass the environment variable to your AI Client:

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

---

🎛 Advanced Usage & CLI Flags

Need granular control? Filter exactly what gets loaded.

# Load everything
uvx --refresh agentic-store-mcp

# Load only specific feature modules
uvx --refresh agentic-store-mcp --modules code
uvx --refresh agentic-store-mcp --modules data memory

# Load only specific isolated tools
uvx --refresh agentic-store-mcp --tools agentic_web_search repo_scanner

# Debug: List what would be loaded and exit
uvx --refresh agentic-store-mcp --list

---

🤝 Contributing

We want to build the ultimate open-source tool directory for Agents!

Tool Guidelines:

• 🐍 Pure Python — No weird managed services, no mandatory accounts.
• 📦 Self-contained — Placed inside agentic_store_mcp/modules/<module>/<submodule>/<tool_name>/.
• 🧪 Tested — Must have a test_handler.py smoke test.
• 📚 Documented — Include a README.md for tool usage.

Local Dev Setup:

git clone https://github.com/agenticstore/agentic-store-mcp
cd agentic-store-mcp
uv sync

# Run tests
uv run pytest

# Test the MCP Server explicitly
uv run server.py

---

📜 License & Support

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

---

⭐ If AgenticStore MCP makes your workflow faster, please star the repository to help others discover it! ⭐

Built with ❤️ by AgenticStore.dev — Open-source AI tooling for everyone.