MCP in production: useful niche, not the glue layer
An Ask HN thread on MCP (Model Context Protocol) in production got a detailed and honest set of responses. The actual use cases were narrow but real: connecting Claude Code to Jira and Figma, bug reporting tools that check for duplicates before filing, enterprise system integrations with Glean and Port. One commenter summed up the drift clearly: 'Many things once needed an MCP server, now could be easily done via using CLI by agent.'
The pattern is that MCP made sense as a standardization layer when agent-to-system communication was awkward. Now that agents are getting better at using CLIs directly, MCP's value proposition narrows to cases where you need structured, persistent, authenticated connections to complex enterprise systems like Jira, where natural language in the terminal genuinely beats the UI.
The counterpoint worth noting: several people said MCP is still their primary way of connecting Cursor to enterprise systems. So it's not dead, it's just settling into a specific niche rather than becoming the universal middleware layer some expected.
So what?
If you're building MCP servers as a product, you're competing with the CLI and losing on simplicity for most use cases. The surviving MCP market is enterprise systems with complex auth, structured data, and UIs bad enough that natural language is a genuine improvement. That's a real but narrow wedge, and founders should be honest about which category they're actually in.