Smithery

Building sophisticated AI agents often involves connecting them to various external tools and data sources. Managing these integrations can quickly become complex, requiring custom code for each connection and leading to fragmented, hard-to-maintain systems. Smithery provides a dedicated platform to streamline this process for developers working with agentic AI. We offer a central registry and hosting solution specifically for agentic services – servers designed to communicate seamlessly with language model agents using the Model Context Protocol (MCP). Our goal is to accelerate the development of agentic AI by making these essential services discoverable, standardized, and easy to deploy.

Key Features

• 🔍 Discover Agentic Services: Find the right MCP servers for your AI applications through our centralized registry. You can browse the platform or programmatically search using the Registry API (GET /servers), leveraging semantic search and specific filters (like owner:, repo:, is:deployed) to pinpoint the services you need. Each listing provides details like descriptions, endpoints, and usage metrics.

• 🚀 Deploy & Host Your MCP Servers: Easily share your own agentic services with the community. Smithery allows you to deploy standard input/output (STDIO) MCP servers, hosting them over a WebSocket connection. This provides benefits like automatic playground generation for user testing, increased visibility in search results, and simplified access for users without requiring local installations. Deployments run in a serverless environment.

• 🧩 Integrate Seamlessly with MCP: Leverage the Model Context Protocol (MCP), an open standard for connecting LLMs with external tools and data. By using MCP servers listed on Smithery, you eliminate the need for custom integration code for each data source or tool. This standardized approach simplifies building and maintaining agents like intelligent IDEs, chat interfaces, and custom AI workflows.

• ⚙️ Standardize Configuration & Connection: Interact with diverse MCP servers through a consistent interface. Smithery utilizes JSON schemas (configSchema) provided by server authors to define the necessary configuration for connection. Whether connecting via WebSocket (wss://server.smithery.ai/${qualifiedName}/ws?config=${base64encode(config)}) or other methods, you'll understand the required parameters upfront.

Use Cases

1. Enhancing an AI Assistant with Real-Time Data: You're developing an AI assistant that needs access to current weather information. Instead of building a custom API wrapper, you search the Smithery registry using the API (GET /servers?q=weather) for an MCP server providing weather data. You retrieve its details (GET /servers/{qualifiedName}), note its required configuration schema, and integrate it into your agent using an MCP client library, connecting via the provided WebSocket URL.

2. Sharing a Custom Internal Tool: Your team has built an internal service that allows an LLM agent to query your company's documentation database via an MCP interface. To make it accessible to other developers and agents within the company, you add the server to Smithery and use the Deployments feature. Smithery hosts the server, provides a stable WebSocket endpoint, and makes it discoverable through the registry, simplifying its adoption across different projects.

3. Building a Multi-Tool Agent Workflow: You are constructing a research agent that needs to fetch academic papers, summarize web content, and manage citations. You use Smithery to discover separate MCP servers specializing in each task (e.g., an arXiv search server, a web scraping/summarization server, a Zotero/bibliography server). Because they all adhere to the MCP standard, integrating these diverse tools into your agent becomes significantly more manageable than dealing with disparate APIs.

Conclusion

Smithery acts as the essential hub for developers building with agentic AI. By providing a centralized registry for discovering MCP servers and offering streamlined deployment and hosting, we simplify the process of connecting your LLM agents to the tools and data they need. Adopting the Model Context Protocol through Smithery means less time spent on boilerplate integration code and more time focused on building intelligent, capable AI applications. Explore the registry to find existing services, or deploy your own and contribute to the growing ecosystem.