This MCP server enables natural language spec description "use speclinter to parse a feature for signing up users" and turn it into a fully realized feature with a user story/epic and its respective tasks.
How does this differ from what Kiro just released? The differentiator is the integration of Gherkin feature files. These files act as the "linter" where the AI system is essentially validating the implementation to the user journey defined.
I found this workflow to be particularly useful because AI is very good at extrapolating from behaviors rather than pure technical spec.
Please give it a try; I'm always open to feedback. I have longterm ideas that I'm holding on to, but plan on releasing an expanded version of this concept in the coming months.
Thanks for reading!