Giving coding agents file outlines to save tokens on ultra-large filesblog.sweep.dev1 points·by williamzeng0·5 miesięcy temu·0 comments
Show HN: Sweep, Open-weights 1.5B model for next-edit autocompletehuggingface.co534 points·by williamzeng0·6 miesięcy temu·153 comments
Using the JetBrains program structure interface for codebase contextblog.sweep.dev3 points·by williamzeng0·8 miesięcy temu·0 comments
Decreasing code editing failures by 38% with output normalizationblog.sweep.dev3 points·by williamzeng0·8 miesięcy temu·0 comments
Show HN: Sweep, AI autocomplete for JetBrains that rewrites codesweep.dev5 points·by williamzeng0·9 miesięcy temu·0 comments
Show HN: Rust BPE tokenizer for Qwen models that's 12x faster than HuggingFacegithub.com5 points·by williamzeng0·10 miesięcy temu·0 comments
Teaching LLMs to spell with token healingblog.sweep.dev2 points·by williamzeng0·10 miesięcy temu·0 comments
Show HN: Building sub-100ms autocompletion for JetBrains IDEsblog.sweep.dev3 points·by williamzeng0·10 miesięcy temu·1 comments
williamzeng0·12 miesięcy temu·discusshttps://sweep.dev/ is better than Copilot/Windsurf in PyCharm (disclaimer I am one of the founders)
williamzeng0·12 miesięcy temu·discussI think the two big things are privacy mode on by default and zero data retention by default.No code or prompts are stored unless you opt-in. We also have on-prem deployment options but it's much more expensive.
williamzeng0·12 miesięcy temu·discussTransparently I'm a founder building a plugin that's like "Cursor for JetBrains IDEs", if you're an windsurf user you might like us: https://plugins.jetbrains.com/plugin/26860-sweep-ai