Replies: 1 comment
-
|
I think a lot of this could be implemented without requiring changes to be made to CodeCompanion. But what I think would really make this feel first-party is if there were good ways to reference back to previously implemented/ worked on features. So maybe something changes about a requirement. If I could select the chat where that was implemented, and just tell it "there were changes, make sure the updated requirement is met", that'd be great. There might also be interesting additional features if using memories, like say it's implementing a task that's tangentially related to a different task, it could then easily look up the memories created by the instance that worked on that |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
First off, huge thanks for building and maintaining CodeCompanion.nvim – it's hands-down the best AI coding experience in Neovim. The agentic workflows, tools, slash commands, and rules make it incredibly powerful and fun to use!
I wanted to suggest a new optional workflow inspired by Amazon's new Kiro IDE, which emphasizes structured planning before coding. Kiro generates specs (requirements, design, tasks) in a dedicated directory, then lets users "start" individual tasks with agents.
It would be amazing if CodeCompanion could add something similar:
:CodeCompanionSpecInit(or via actions palette).codecompanion_spec/(or configurable):requirements.mddesign.mdtasks.md(with checkbox tasks, dependencies noted)Then, in any chat:
/spec-task 3to focus the chat on implementing a specific task (pre-populating the prompt with the task description, design, and requirements)This would fit perfectly with existing agentic workflows – users could approve diffs per task, keeping everything structured without leaving Neovim.
Here are short examples of what the generated files could look like for a simple feature ("Add dark mode toggle to a Lua config"):
requirements.md
Beta Was this translation helpful? Give feedback.
All reactions