Log inSubmit server
AgenticStore logo

AgenticStore

Local stdio MCP toolkit for prompt privacy, code analysis, GitHub workflows, security checks, data inspection, and SearXNG search.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install AgenticStore

Pick your client, then add it in one click or paste the config.

STDIOPython
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "agentic-store-mcp": {
      "command": "uvx",
      "args": [
        "agentic-store-mcp"
      ]
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Install not yet sandbox-tested— the server itself is confirmed live (we called its tools/list endpoint). The stdio install command is a separate, ongoing check.

Overview

AgenticStore MCP server runs locally over stdio and gives AI clients a broad toolkit for software development, security review, data inspection, and web research. Its capabilities include a prompt firewall for PII and secret sanitization, persistent memory, audit recording, local code analysis, GitHub operations, and self-hosted SearXNG search. Reach for it when an agent needs bounded access to local repositories and files, selected GitHub workflows, or privacy controls before prompts leave the machine. The firewall is tested and stable on macOS, while Linux and Windows support is planned.

Use cases

  • Review local commit history and changed symbols
  • Scan a project for leaked secrets and vulnerable dependencies
  • Search GitHub repositories and manage issues or pull requests
  • Profile and validate local structured data
  • Search the web through a self-hosted SearXNG instance

Key features

  • Local MCP stdio server
  • Prompt firewall with PII and secret sanitization
  • Persistent memory and token optimization
  • Local code, dependency, and repository analysis
  • GitHub metadata, issue, pull request, and security tools
  • Self-hosted SearXNG web search

Tools74

What AgenticStore lets your agent do. Expand a tool for its parameters.

Verified live~16.1k tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked 2d ago

Recent health check succeeded.

5/5 checks healthy over the last 19d

What it can do

From the 74 tools this server returned when we ran it; 74 declare their own safety hints.

Runs commands · 3Can delete or modify data · 9Handles credentials · 20Makes network requests · 12Read-only tools · 63
Unusual tool instructions. 2 tool descriptions contain text aimed at the AI model rather than at you (reaches for local secrets). Read the tool descriptions before connecting.
Network at startup2 connections
Files written at startupNone

Observed 2026-10-27 in an isolated sandbox. A description of reach, not a security audit.

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 2d ago via OSV.dev · agentic-store-mcp (PyPI)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars5Stargazers on the source repository.
Last commit13d agoMost recent push to the default branch.
Tools exposed74Callable tools this server registers over MCP.
Directory activity1 viewsConfig copies, upvotes, and views on AllMCPs.

How AgenticStore works

What AgenticStore does

AgenticStore combines an MCP tool collection with local controls for handling prompts and project data. It is designed for AI-assisted development workflows that need repository analysis, security checks, controlled file and data operations, GitHub collaboration, and web research from one local service.

The project describes 74 specialized tools, including code structure and reference analysis, dependency auditing, secret and PII scanning, release checks, bounded test execution, archive inspection, data profiling, contract validation, and local document comparison. Its GitHub tools can inspect repository metadata, fetch known files, search code, manage issues, open pull requests, and retrieve code-scanning or Dependabot alerts.

A prompt firewall can intercept traffic through a local proxy, record sanitized prompts, apply rule-based redaction, and optionally use a local Ollama model for sanitization. Persistent memory, token optimization, context pruning, and the Relay project-delivery workflow are also part of the broader toolkit.

How it works

The MCP server runs on the local machine and exposes tools to connected AI clients. Local-only operations read repositories, source files, dependency manifests, SQLite databases, archives, and structured data without using network access. Several operations are deliberately bounded: code symbol scanning covers up to 2,000 files, data profiling covers up to 10,000 rows, archive inspection returns up to 5,000 entries, and read-only database queries return at most 500 rows.

Remote GitHub operations use the GitHub API. Public repository metadata, file retrieval, and code search can work without a token, while private repositories and write or security operations require appropriate GitHub credentials. Web search uses a self-hosted SearXNG instance, which must be running locally. The firewall is configured through the local GUI, including certificate installation and proxy startup.

Setup and configuration

Python 3.12 or newer is supported. Installation can use uvx or pip, and the project also documents running from GitHub source. MCP clients can be configured manually with JSON or through the local GUI dashboard.

Start SearXNG separately with the project’s Docker Compose setup before using agentic_web_search. Connect Ollama only when local-model prompt sanitization is needed. The firewall setup is available through the GUI rather than the basic MCP configuration flow.

GitHub features that access private repositories or perform changes require a GitHub token. Pull requests and issue mutations first return a preview and require an explicit confirmation before execution. Local test, lint, build, or typecheck execution likewise requires confirm_execute=true.

Tools and capabilities

  • Analyze local Git history, changes, symbols, references, dependencies, and release readiness.
  • Search local directories or GitHub code and retrieve known files from GitHub repositories.
  • Inspect Python quality, dependency vulnerabilities, leaked secrets, PII, and GitHub security alerts.
  • Run bounded read-only checks over CSV, JSON, JSONL, SQLite, archives, documents, and API contracts.
  • Transform selected local CSV or JSONL data into a new file after confirmation.
  • Query local SQLite databases with restricted read-only SQL.
  • Search the web through SearXNG and extract page metadata and content from URLs.
  • Manage GitHub issues and pull requests with preview-and-confirm safeguards.

Limitations and notes

The firewall currently has stable tested support on macOS; Linux and Windows support is planned. Local code, repository, and data tools are read-only unless a capability explicitly creates a new output or executes a confirmed operation. The server does not replace remote Git operations: local Git analysis and remote GitHub API actions are separate tool scopes.

SearXNG is a required local dependency for web search. GitHub security alerts need tokens with security_events or repo scope, and GitHub write operations need a token with repo scope. The prompt firewall is intended to control traffic from supported client workflows; enabling it requires the local GUI and certificate setup.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about AgenticStore

AgenticStore is an MCP server that connects AI clients to local files, Git repositories, GitHub, and optional self-hosted SearXNG and Ollama services. Its main tools cover code analysis, dependency and secret scanning, data inspection, web search, GitHub operations, prompt sanitization, audit recording, and persistent agent memory.

Install AgenticStore with `uvx` or `pip`, then add it to an MCP client manually or through its local GUI. Python 3.12 or newer is supported, and the project also documents running from GitHub source.

Yes, AgenticStore supports Cursor. The server is a local stdio MCP server, and its client setup can be handled through manual JSON configuration or the local GUI.

Run a self-hosted SearXNG instance locally before calling `agentic_web_search`. The documented startup path uses Docker Compose, and the search tool returns ranked results with titles, URLs, and snippets.

Yes, private-repository access requires a GitHub token. Read operations such as repository metadata and file retrieval use the GitHub connector, while issue, pull request, and security operations require the token scopes documented for those actions.

Related MCP servers

All in Developer Tools All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewAgenticStore AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/agentic-store-mcp?style=directory)](https://allmcps.com/mcp/agentic-store-mcp)
HTML Embed
<a href="https://allmcps.com/mcp/agentic-store-mcp"><img src="https://allmcps.com/api/badge/agentic-store-mcp?style=directory" alt="AgenticStore on AllMCPs" /></a>