backseat-pilot

vscode extension for using local LLMs with vscode

server

based on the llama-cpp-python server example.

extension

LLM url is configuraable:

"backseat-pilot.url": "http://localhost:8000/v1/completions",
"backseat-pilot.maxTokens": "512"

Publish your own version

npx vsce package

usage

Commands based on ai_extension_vscode.

cmd shift p => LLM Chat, LLM Refactor, LLM Description 2 Code