Stage 2: Capture Intent
Duration: ~3 minutes
What happens: You tell AgentSherpa what you want to build. The agent confirms its understanding before doing anything.
How it works
AgentSherpa asks: "What kind of MCP server do you want to build?"
You describe your idea in plain language. The default workshop project is a stock market MCP server that fetches live data from the Yahoo Finance API, but you can describe your own idea.
AgentSherpa then mirrors back what it heard:
-
The server’s purpose
-
The data source
-
The type of tools it will expose
-
Who will use it
The agent asks you to confirm or correct before proceeding.
Why this matters
In the agentic SDLC, intent capture replaces traditional requirements gathering. Instead of writing a spec document, you describe what you want in conversation. The AI agent confirms understanding and uses this intent as the foundation for every later stage.
This is the "contract" between you and the agent — it will reference your stated intent throughout the build.