Smitherys role in solving MCP challenges

From: aidotengineer

Smithery’s Role in Solving Model Context Protocol Challenges

Smithery, founded by Henry, aims to address significant problems within the Model Context Protocol (MCP) space and ecosystem [00:00:14]. Henry, who is also a member of the MCP committee, launched Smithery in December 2024 to tackle these emerging issues [00:04:58].

The Genesis of Smithery

The founding of Smithery stems from the observation of “Claude’s Paradox” [00:01:45]. Despite the significant intelligence developed by frontier labs, this intelligence often remains “stuck in a box” [00:01:52]. To make AI agents practically useful, there’s a need to consider context and capability – what inputs and outputs models require [00:01:55].

This problem was recognized by major labs, leading Anthropic to release the Model Context Protocol (MCP) in November 2024 [00:02:08]. The MCP was designed as an open standard to help LLMs connect to various services, promising to standardize the “n times n problem” of connections [00:02:11]. While the emergence of MCPs and a vibrant developer community was exciting, it also brought a new set of challenges [00:02:22].

Addressing MCP Ecosystem Challenges

Smithery aims to become an “AI gateway” that can grow and orchestrate the new era of AI-native services for AI agents [00:05:06]. It addresses problems in the MCP ecosystem for both users and developers [00:04:51].

For MCP Users:

• Fragmentation: The increasing number of MCP servers makes it difficult to find high-quality ones [00:02:48]. Smithery plans to curate thousands of MCPs to address this [00:05:28].
• High Friction Install: The typical five-step installation process for MCPs on GitHub is cumbersome, and insecure MCPs can be installed [00:03:14].
• AI Native Economy: There is no clear plan for creating an economy of AI-native services or handling agentic payments, raising questions about how agents will pay on behalf of users or how to avoid numerous small subscriptions [00:03:30].

For MCP Developers:

• Hosting Problems: While Streambo HTTP transport has made hosting easier, developers still face issues with stable sessions and resumability [00:03:50].
• Lacking Developer Tooling: Current tools are basic, leaving developers with questions about designing the best MCPs, ensuring their tools are called, and creating optimal agent experiences [00:04:06].
• Distribution: Developers lack clear pathways for their MCPs to be discovered [00:04:30].
• Observability: Improving MCPs after deployment is challenging [00:04:36].
• Monetization: There is no clear strategy for developers to monetize their MCPs [00:04:43].

Smithery Playground Demo

Smithery has developed a playground to showcase what an AI agent can achieve when it has access to thousands of curated MCPs [00:05:22]. In a demonstration, an AI agent successfully:

1. Analyzed a prompt to find the most pressing issue on a GitHub repository [00:05:33].
2. Used a search servers function to find and connect to relevant MCPs, including GitHub and Linear [00:05:51].
3. Utilized the GitHub MCP to identify high-priority bugs [00:06:06].
4. Created a detailed ticket on Linear, including a link to the original issue [00:06:17].

This end-to-end task demonstrates the potential when an AI agent can seamlessly connect to multiple MCPs [00:06:40].

Future Vision

Henry believes that the future of the internet will be dominated by “tool calls” rather than clicks, where the “agent experience” will supersede the user experience [00:07:09]. This future will be built collaboratively by the entire developer community [00:07:21].