Repository navigation
SKonsole: Streamlining PR and Commit Message Authoring with Semantic Kernel #106
Closed
Lee Miller (lemillermicrosoft)
started this conversation in
Show and tell
Replies: 3 comments 2 replies
|
I love the idea and will try using it in my side project. Thank you for your awesome code, Lee Miller (@lemillermicrosoft) ! |
0 replies
|
Great idea Lee. Along similar lines I would love to have an AI code reviewer that I can do a pre-review with before I send a broader code review to my crew. It would handle coding conventions, compliance related policies, SAL annotations etc. Have you put some thought into this type of feature set? This is especially powerful on larger organizations and code bases where conventions can differ across the source tree, policies are updated on a cadence etc. |
1 reply
|
I've updated the tool and made it available for installation via dotnet tool install --global SKonsole |
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Are you tired of spending too much time crafting the perfect commit message or PR feedback? SKonsole might just be the tool you need. This console app, built on Semantic Kernel, uses various skills to generate commit messages and pull request feedback.
The SKonsole repository contains projects and classes that help automate the process of authoring pull requests and commit messages. The
PRSkill, for example, can generate feedback, commit messages, and pull request descriptions based ongit difforgit showoutput. This skill is built on top of theCondenseSkill, which is a skill that can condense multiple chunks of text into a single chunk.The app requires the Semantic Kernel as a dependency and can be found on the GitHub repository: https://github.com/lemillermicrosoft/skonsole. While the app currently supports only two skills, the repository includes instructions on how to expand the app's capabilities.
Note that this is just the beginning of my exploration of the Semantic Kernel's capabilities. In the future, I plan to add more features such as bot chat, planning, semantic memory, code generation, and more. Stay tuned for updates!
All reactions