I'm with Linus
Linus Torvalds advocates for viewing AI as a useful tool rather than banning it, emphasizing understanding and control over its output. The debate around AI reflects broader societal polarization and resistance to change. (isene.org)
Show HN: Yana – A local-first, human-governed AI runtime and IDE in Rust
github.com5 points0 comments
Yana is a local-first, cross-platform control plane and IDE that mediates what AI models can propose and actually do, implemented primarily in Rust. It enforces a deterministic pipeline - authority (who may ask), capability (what kind of power), policy (rules then human approval), and executor (perform only permitted operations) - so a model may propose actions but Yana decides and a human can always stop execution. The system combines a Rust runtime (yana-rt) with governance features: capability manifests, human approvals, a hash-chained audit log, a core-lock SHA-256 manifest that pins core files, and an independent watcher that can halt tool calls. Yana ships adapters for many providers (19 total across cloud and local) and emphasizes that skills teach agents without granting permissions; only capabilities can grant action rights.
The desktop Studio workspace integrates files, a CodeMirror editor, PTY terminals, chat (streamed through yana-rt), a guarded design canvas, task management requiring evidence, and governance telemetry showing allowed/blocked activity. The codebase lists extensive components and counts (66 hooks, 71 rules, ~100 agents, 2,026 skills, 170 commands), a 40-subcommand runtime, and no Python runtime dependency. Builds and installers are provided for macOS, Windows, and x86_64 Linux (no Linux ARM release), with local model support for runtimes like Ollama, llama.cpp, LM Studio and others.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
More in AI
Publish fake stats, get cited in LLMs
A researcher created a website with 50,000 fake statistics to test how AI models like ChatGPT incorporate fabricated data. The fake stats were cited in over 10,000 AI sessions, academic papers, and by major companies, highlighting vulnerabilities in AI reliance on unverified sources. (peec.ai)
Show HN: Make AI coding agents verify its own work
Reticle is an AI tool that enables coding agents to verify their own work, improving accuracy and reliability. It integrates with web and desktop applications to enhance AI-generated code testing and validation. (github.com)
DeepSeek just open-sourced its inference engine
DeepSeek has open-sourced DeepInfer, an inference engine designed for high-performance, large-scale deployment of DeepSeek models. The framework is available on GitHub with 3,100 stars and supports efficient AI inference tasks. (github.com)
Ask HN: How to get the most intelligence out of your agents?
ChatGPT offers superior insights compared to Codex but has limited tool calling capabilities, according to a user. They want to combine both models' strengths locally on their laptop for better AI agent performance. (news.ycombinator.com)
Parallel Language Models via Variational Decoding
Parallel Language Models (PLMs) generate multiple tokens simultaneously using latent auxiliary variables, reducing inference latency compared to traditional autoregressive models. Empirical results show that PLMs achieve faster inference speeds with comparable or better performance across various benchmarks, especially at larger scales. (arxiv.org)