devtake.dev
AI Security Hardware Open Source Policy Apple

#vibe-coding

RSS
Related: #ai-models · 1 #benchmarks · 1 #bun · 1 #delegation · 1 #dev-tools · 1 #javascript · 1 #llm · 1 #microsoft · 1 #open-source · 1 #runtime · 1
Figure from a forensic document-examination study used here as visual shorthand for verifying authorship of code.
Open Source·

yt-dlp's maintainer says Bun is now 'fully vibe-coded'. Support is officially deprecated.

yt-dlp's maintainer bashonly says Bun's Rust rewrite 'has taken a turn towards being fully vibe-coded.' The supported window narrowed to four versions.

The DELEGATE-52 project repository on GitHub, showing Microsoft's benchmark for testing LLM document editing fidelity
AI·

Microsoft tested 19 LLMs as document editors. Even the best ones corrupted 25% of the content.

The DELEGATE-52 benchmark tests AI editing across 52 professional domains. Frontier models corrupt a quarter of document content over long workflows.