DOCSEARCHbyAlgolia
Submit Docs

Awesome MCP DevTools

repository·main·Indexed 19 days ago

https://github.com/punkpeye/awesome-mcp-devtools

A curated directory of tools, SDKs, and resources for developers working with the Model Context Protocol (MCP). It includes language-specific SDKs (TypeScript, Python, Go, Ruby, Swift, etc.), high-level frameworks for server scaffolding and orchestration, testing and debugging utilities like the official MCP Inspector, as well as proxies, gateways, and deployment resources for MCP servers and clients.

Tokens
2.8K
Snippets
0
Records
12
Agent score
17%

What's inside awesome-mcp-devtools

  1. Overview of Awesome MCP DevTools

    main

    Awesome MCP DevTools is a curated collection of developer tools, SDKs, libraries, utilities, and resources specifically designed for working with the Model Context Protocol (MCP). It serves as a central directory for developers looking to build, test, or deploy MCP servers and clients across various programming languages.

    Key categories include:

    • SDKs: Language-specific kits for server and client development.
    • Frameworks: High-level abstractions for building MCP-enabled applications.
    • Testing Tools: Utilities for verifying MCP functionality, including authorization testing.
    • Libraries & Utilities: Proxies, gateways, and development tools.
    • Hosting & Templates: Resources for deploying and scaffolding MCP projects.

  2. High-level MCP Frameworks

    main

    The following frameworks provide high-level abstractions for building, orchestrating, or scaffolding Model Context Protocol (MCP) servers:

    Server Scaffolding & Creation

    • mcp-craft: CLI toolkit for scaffolding, testing, and registering MCP servers across Claude, Cursor, VS Code, Codex, and Gemini.
    • template-mcp-server: A CLI tool to create a new MCP server project with TypeScript support.
    • create-mcp-ts: Creates a new MCP server in TypeScript with 'batteries included' and supports user-defined templates.
    • xmcp: A TypeScript framework featuring file-system routing and a complete toolkit.

    Orchestration & Agent Frameworks

    • cordum: A native MCP server from a safety-first agent orchestration platform. Supports stdio + HTTP/SSE transports and includes tools for job submission, policy evaluation, and workflow triggering.
    • hasmcp-ce: A no-code, self-hosted bridge that converts APIs to MCP servers with built-in authentication, authorization, and real-time logging/metrics.
    • mcp-agent: Provides simple, composable patterns for building effective agents using MCP servers.
    • FEGIS: A semantic programming framework that compiles YAML archetypes into structured tools with built-in memory and an emergent knowledge graph.
    • mcp-mesh: A distributed orchestration framework for MCP servers featuring automatic discovery, dependency injection, and Kubernetes-native scaling.
    • semantic-kernel: Microsoft's enterprise-ready orchestration framework, compatible with MCP for building multi-agent systems.
    • gpt-computer-assistant: A framework designed to build vertical AI agents.

    Specialized Frameworks

    • photon: A runtime and marketplace for MCP servers using a single-file TypeScript format (similar to npm for MCP).
    • framework-rai-mcp: Focused on responsible AI and ethical deployment for startups and enterprises.
    • solana-agent-kit: An MCP SDK specifically for Solana.
    • Kryfto: An open-source web-browsing backend providing a 42-tool MCP server for Claude Code, Cursor, and Codex.

  3. Understand the Legend and Resource Markers

    main

    The repository uses specific icons to denote the nature and language of the listed resources:

    Resource Types:

    • 🎖️: Official MCP resource

    Programming Languages:

    • 📇: TypeScript
    • 🐍: Python
    • ☕: Java
    • 🏎️: Go
    • 🦀: Rust
    • 🔶: Kotlin
    • #️⃣: C#
    • 〽️: Scala
    • 🎯: Dart
    • 💎: Ruby
    • 💧: Elixir
    • 🌊: C/C++
    • 🍎: Swift
    • 👾: Bash
    • λ: Common Lisp

  4. Use MCP Proxies and Gateways to bridge transports or manage access

    main

    MCP Proxies and Gateways allow you to bridge different transport protocols (e.g., converting stdio to HTTP SSE), route traffic to multiple backend servers through a single endpoint, or add security layers.

    Key capabilities include:

    • Transport Conversion: Convert stdio-based MCP servers to HTTP SSE (e.g., boilingdata/mcp-server-and-gw, sparfenyuk/mcp-proxy, punkpeye/mcp-proxy).
    • Cloud Access: Enable cloud-based AI services to access local stdio servers via HTTP/HTTPS (e.g., EvalsOne/MCP-Connect).
    • Multi-Server Routing: Use a single endpoint to route requests to multiple backend MCP servers (e.g., mcpproxy-go, multi-mcp, TBXark/mcp-proxy).
    • Security & Governance: Add policy-based access control, human approval workflows, or tool filtering (e.g., sidclawhq/platform, mcpproxy-go).
    • Network Traversal: Use gateways with NAT traversal to provide MCP services across networks (e.g., fangyinc/mcpport).

  5. Host and Deploy MCP Servers

    main

    If you need to host MCP servers at scale or as a service, consider these options:

    • Managed Services:
      • Glama: Discover and manage open-source MCP servers.
      • Smithery: Cloud hosting for MCP servers via Docker containers.
    • Infrastructure/Orchestration:
      • vitorbari/mcp-operator: A Kubernetes operator that provides automatic protocol validation, horizontal scaling, and observability for MCP servers.

  6. Use MCP Templates to bootstrap new projects

    main

    Start your MCP development with pre-configured templates:

    • TypeScript/FastMCP: fastmcp-boilerplate (includes TypeScript, ESLint, and Prettier).
    • Dart: dart-mcp-server-template (includes Docker configuration and http+stdio transport bindings).
    • Ruby on Rails: rails-mcp-startup-boilerplate (includes Rails 8.0.2, Devise, Doorkeeper, FastMCP, and Stripe integration for paid MCP servers).

  7. Testing and Debugging MCP Servers

    main

    Use these tools to verify, debug, and benchmark your MCP implementations:

    Inspection & Debugging

    • modelcontextprotocol/inspector: The official UI for testing MCP servers.
    • mcp-cli: A command-line inspector for manual testing.
    • mcpjam/inspector: General testing and debugging tool.
    • muppet-kit/inspector: MCP Inspector with AI-assisted debugging capabilities.
    • Companion: A utility for testing and debugging MCP servers on macOS, iOS, and visionOS.
    • mcp-doctor: A zero-config CLI that auto-discovers MCP configs (Claude Code, Cursor, VS Code, etc.), tests connections via JSON-RPC handshake, and audits security/latency.

    Quality, Security & Evals

    • AgentTrust: Quality verification using LLM judge consensus and adversarial probes.
    • mcp-evals: A package and GitHub Action for running evaluations.
    • aguara: A static security scanner that detects prompt injection, credential leaks, and taint tracking in configs and tool descriptions.
    • mcp-chaos-rig: A local MCP server that breaks on demand (auth failures, flaky responses, etc.) to test client resilience.
    • mcp-observatory: A CLI and MCP server that provides health scoring, schema quality audits, and protocol conformance checks.
    • mcp-vet: A zero-config CLI that scans source code (TS/JS/Python) for patterns that break the 2026-07-28 MCP spec.
    • veridict: Runtime trust scoring middleware that classifies failures and produces a trust verdict (yes/caution/no).

    Performance & Load Testing

    • MCPLoadTester: A JMeter sampler for load testing MCP servers over stdio, HTTP, and SSE.
    • mcpbr: A benchmark runner for evaluating performance and agentic capabilities.

  8. Use Development Tools to inspect, test, and debug MCP servers

    main

    A variety of tools exist to assist in the development lifecycle of MCP servers:

    Inspection and Interaction

    • CLI Inspection: Use f/MCPTools to inspect and interact with servers, or flux159/mcp-chat for a CLI-based chat client.
    • Logging & Auditing: Use ithena-cli to log interactions locally for debugging, or mcptee to proxy MCP and log inputs/outputs to a YAML file.
    • Shell Integration: Use inercia/mcpshell to turn shell scripts into MCP tools, or rodaddy/mcp2cli to wrap MCP servers as bash-invokable commands.

    Testing and Automation

    • Autotesting: Use strowk/mcp-autotest to run YAML-based, language-agnostic autotests.
    • Hot Reloading: Use strowk/synf to hot-reload an MCP server when its source files change.

    Code Generation

    • Spec-to-MCP: Transform OpenAPI specs into MCP servers using agentify or taskade/mcp. Use type-mcp/mcp-anything to auto-generate servers from various codebases (FastAPI, Express, Go, etc.).

  9. Testing MCP Authorization Flows

    main

    To test MCP servers that require authentication and authorization, you can use the following resources:

    Implementation Examples

    • http-oauth-mcp-server: An example implementation of a remote MCP server using OAuth.
    • modelcontextprotocol/python-sdk: An example of an authenticated SSE server provided in the official Python SDK.

    Public Test Endpoints

    You can use these production SSE endpoints to test your OAuth flows:

    • Asana MCP Server: https://mcp.asana.com/sse
    • Sentry MCP Server: https://mcp.sentry.dev/sse
    • Atlassian MCP Server: https://mcp.atlassian.com/v1/sse (Note: requires allowlisting)

  10. Explore MCP SDKs by Language

    main

    The project categorizes SDKs (Software Development Kits) to help you find the right tool for your stack. SDKs are ordered by popularity (GitHub stars).

    JavaScript/TypeScript SDKs

    • FastMCP: High-level framework for TypeScript.
    • QuantGeekDev/mcp-framework: Fast and elegant TypeScript framework.
    • wong2/LiteMCP: High-level framework for JS/TS.
    • ModelFetch: Runtime-agnostic SDK for JS/TS environments.
    • NeuroLink: Multi-provider AI SDK with built-in MCP client support.
    • FrontMCP: TypeScript-first framework using decorators and dependency injection.

    Python SDKs

    • FastMCP: High-level framework for Python.
    • mcp-use: Library to connect LLMs to MCP servers (local/remote).
    • langchain-mcp: Provides MCP tool calling support for LangChain.
    • lc2mcp: Converts LangChain tools to FastMCP tools.
    • easymcp: Asyncio native client SDK with namespacing and caching.
    • mcp-cli: Lightweight CLI client for remote MCP servers.

    Go SDKs

    • modelcontextprotocol/go-sdk: 🎖️ The official Go SDK, maintained with Google.
    • mark3labs/mcp-go: Golang SDK for servers and clients.
    • metoro-io/mcp-golang: Type-safe Golang framework.

    Other Notable SDKs

  12. MCP Reusable Libraries and Components

    main

    These libraries provide reusable logic to extend MCP functionality:

    Tooling & Integration

    • codemode: Programmatically turns any OpenAPI spec into two sandboxed tools (search + execute) instead of writing individual tools.
    • mcp-langchain-ts-client: Allows using MCP-provided tools within LangChain.js.
    • express-mcp-handler: Binds an MCP server to an Express server using the StreamableHTTP Transport.
    • mcp_auto_register: A Python tool to automate the registration of functions and classes from a package into a FastMCP instance.

    Observability & Payments

    • openllmetry (MCP instrumentation): OpenTelemetry instrumentation for MCP Python to capture tool calls, notifications, and traces.
    • mcp-otel-go: OpenTelemetry instrumentation for Go MCP servers using the official go-sdk.
    • monapi: Adds per-tool x402 micropayments to any MCP server, allowing agents to pay in USDC per tool call.