Better Language Server Tools for LLMs
What happened
The article discusses how language server tools, which provide semantic information, diagnostics, and refactoring capabilities for code, can be adapted for large language models (LLMs). It notes that LLMs lack cursor-based positioning and work best with pure text, suggesting a redesign of tool formats to match this modality. The author compares existing LSP calls that use line and character indexes with token‑based inputs for LLMs, highlighting the inefficiency of current approaches. An example of OpenAI’s Codex apply_patch format is given as a model for a more LLM‑friendly query language. The piece proposes creating a language‑server‑style interface that aligns with LLM token streams and post‑training.
Why it matters
If tools can give LLMs fast, token‑efficient code information, AI coding assistants could become quicker and cheaper to run, benefiting developers who rely on them. This could also make AI‑driven code fixes more reliable for everyday programming tasks.
Reality check
An automated, sceptical read of the sourceIncremental. The ideas build on existing language server concepts and OpenAI’s patch format rather than introducing a wholly new technology.
Is this a big deal?
Get stories like this every morning
The day's AI news in plain English, in your inbox at 8:30 IST. Free.
More AI news
Every Way to Get 192GB for Local AI (Without a $9,500 Mac) [video]Industry news · 14 Nov 2026DeepSeek V4 Pro 0813DeepSeek · 14 Nov 2026DeepSeek V4 Pro 0813DeepSeek · 14 Nov 2026Show HN: Portal2Bench – Can AI agents solve Portal 2?Industry news · 14 Nov 2026Anthropic Upgrades Claude Code Projects to Run Parallel AI Workers Hands-FreeAnthropic · 14 Nov 2026This page was collected from michaelorenstein.com and summarised automatically by AIPulsee's pipeline. See how it runs.