AI cannot replace what KCL compilers or Kusion engines do.
AI cannot replace the value of collaborative programming and scalability in KCL, but we can further improve the efficiency of individuals and organizations by using AI to share and reuse KCL knowledge as much as possible.
We cannot use the existing LLM to help write KCL code because they do not contain knowledge of KCL and require precise tuning using KCL data.
Goals
We can use the vscode-kcl extension to ask any questions about kcl codes and documents (A dialog or copilot-like capability). For example, a dialog with the hint enter the kcl promot.
Design
API
Ping
Chat
POST /v1/chat/completions
req body: { message: Prompt }
res body: { completions: Completion[]}
Enhancement
Add Code and Documentation AI Assistants for KCL
Why
Goals
enter the kcl promot
.Design
API
Ping
Chat
KclService
https://kcl-lang.io/docs/reference/xlang-api/rest-api
CLI
IDE Integration
cmd + shift + p
=>enter the kcl promot
Impl
Next.js + VSCode Plugin
Reference