Smithery

Smithery is a comprehensive platform for connecting AI agents to tools and services through the Model Context Protocol (MCP). It provides a registry of over 715 MCP servers, handling authentication, credential management, and session persistence so agents can securely interact with external services without manual configuration.

Available at smithery.ai, the platform serves both as a marketplace for discovering MCP servers and as a managed infrastructure layer that abstracts away the complexity of connecting agents to third-party APIs, databases, and services.

The Model Context Protocol (MCP) is an open standard developed by Anthropic that defines how AI agents connect to external tools and data sources. Smithery builds on MCP by providing a hosted registry, runtime, and management layer on top of the protocol.

Popular MCP servers available on Smithery include web search engines (Exa Search, Tavily, Keenable), messaging platforms, cloud services, developer tools, and data sources. Each server exposes a set of tools that agents can discover and invoke dynamically, with Smithery managing the authentication and connectivity lifecycle.

• Managed Auth - Smithery handles OAuth flows, API keys, and session tokens so agents don't need to manage credentials.
• Server Registry - Browse and discover hundreds of verified MCP servers with usage stats and documentation.
• Toolbox - Build custom collections of MCP servers for specific workflows or agent personas.
• Skills - Reusable agent capabilities built on top of MCP servers, such as web research or code analysis.
• Usage Analytics - Track which MCP servers your agents use most and monitor reliability.
• Verified Servers - Curated and tested MCP servers with quality guarantees and community ratings.

Smithery Skills are higher-level agent capabilities that combine one or more MCP servers with curated prompts and configuration presets. For example, a "Web Research" skill might combine Exa Search, web scraping, and content extraction servers with prompts optimized for research workflows.

Skills make it easy to equip agents with complex capabilities without manually wiring up individual MCP servers. They are shareable and can be published to the Smithery community marketplace, enabling teams to standardize agent capabilities across their organization.

Developers and service providers can publish their own MCP servers to the Smithery registry. The platform supports both stdio-based (local subprocess) and HTTP-based MCP servers. Published servers can include custom authentication flows, usage documentation, and example tool calls.

The "Publish" workflow guides developers through server configuration, testing, and verification. Once published, servers are available to all Smithery-connected agents and can be added to user toolboxes with a single click.

Smithery offers a free tier for individual developers to discover and use MCP servers. Paid plans provide additional features like advanced analytics, team management, custom server hosting, and priority support. Detailed pricing is available at smithery.ai/pricing.