Extension implementation of projection boxes, kind of. Calling on an LLM to display relevant variable and function information.
- VS Code version ^1.108.1
- node v20
- OpenAI API Key
- clone repository
- set API via VS Code settings JSON
- open command palette: mac
Ctrl+Shift+P - type and select "Preferences: Open User Settings (JSON)"
- add this line anywhere in settings.json (before the closing }):
"pbExtension.openaiApiKey": "your-api-key-here" - replace
your-api-keywith your actual key - add this line too:
"pbExtension.llmFilteringEnabled": true - save the file
npm installfrom rootnpm run compilefrom root- F5 to launch extension
Pre-req: make sure that the test file is executable and correct
- In the popped up extension window open the test directory and one of the test Python files.
- Open the command palette and run "Test Debug Executor" command
Expected:
- The file will execute with the debugger running
- You will see per line traces as the debugger steps through the file
- Result of executable file should show up on the terminal
- After execution, there will be a slight delay of values displays
- LLM filtered output will be an in-line ghost text decoration (git blame style)
- Hovering over the line will show the full trace using codelens