-
Notifications
You must be signed in to change notification settings - Fork 172
Description
Description
Hi,
As I am trying to get my head around writing custom agents for our organisation, the modernize-dotnet agent is great for seeing some techniques and patterns. It would be great if you have the time to share some insights on how you approached designing + building the agent. Also how you tested (I know it's a fuzzy field).
On thing I instantly asked myself is your choice of inline? MCP server:
mcp-servers:
Modernization:
type: 'local'
command: 'dnx'
args: [
'Microsoft.GitHubCopilot.Modernization.Mcp',
'--prerelease',
'--yes'
]
tools: ['*']
env:
APPMOD_CALLER_TYPE: copilot-cliCould you explain a bit what is going on here? Is there some reference docs I can read up?
What I assume you do:
You locally start the Modernization.Mcp binary which has the actual modernization logic (the classic "do the work" code) plus an MCP server on top.
Why did you choose to use MCP here and not a single binary e.g. "dnx Modernization "
Again, it's a whole new world and I just try to learn so I'd love to understand some of your design choices.
Thank you