From: aidotengineer
The Model Context Protocol (mCP) ecosystem, while exciting, presents several challenges specifically for developers [00:04:47]. These issues range from infrastructure concerns to tooling and economic models.
Hosting Challenges
Despite advancements like Streambo HTTP transport making it easier to find hosting platforms [00:03:55], hosting MCPs remains a significant challenge [00:03:59]. Developers must contend with issues such as stable sessions and resumability [00:04:01].
Lacking Developer Tooling
Developer tooling in the MCP space is currently quite limited [00:04:06]. While a basic MCP inspector is provided by the official MCP repository for testing tools and checking prompts [00:04:10], many open questions persist for developers, including:
- How to design the most effective MCPs [00:04:21].
- How to predict if a specific tool will be utilized by an agent [00:04:24].
- How to create the optimal agent experience [00:04:27].
Distribution Issues
A key problem for developers is the distribution of their MCPs [00:04:30]. There is a lack of clear mechanisms for an MCP to be discovered once it’s created [00:04:32].
Observability
Once an MCP is deployed and in use, developers face difficulties in improving it due to limited observability [00:04:35].
Monetization
A significant unanswered question for developers is how to generate revenue from their MCPs [00:04:43].
Addressing the Challenges
Smithery was founded in December 2024 to address these challenges in the MCP ecosystem [00:05:01]. The company aims to become an “AI gateway” to foster and orchestrate the growing landscape of AI-native services for AI agents [00:05:06]. The ultimate goal is to enable an agent experience where AI agents can access and utilize thousands of curated MCPs [00:05:28]. The future internet is predicted to be dominated by tool calls rather than clicks, emphasizing the importance of the agent experience [00:07:09].