Claude Code added LSP tool for code intelligence features(github.com)
github.com
Claude Code added LSP tool for code intelligence features
https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md
2 コメント
Some more discussion: https://news.ycombinator.com/item?id=46355165
One of the reasons an LSP is so useful is that it does not fabricate anything. If my LSP can't tell me something, it means it doesn't exist or I haven't defined it.
In some ways it helps against code being too magic as well, if my LSP can't understand what's happening in the code it is probably because of a magic string, hidden reference or hidden dependency. If you're in a loosely typed language, it also helps spot when you've got some type shenanigans happening.
All of this is instantaneous, and always factual, deterministic, sourced from hard truth.
I do not know how this LLM LSP functions, but instant, always factual and deterministic are not features of LLMs typically.
I'll give it a go though, like I said I don't know what it does under the hood.