Launching today

Lugh
The programming language for AI agents that ask first
2 followers
The programming language for AI agents that ask first
2 followers
Lugh lets an AI write the program and makes it ask before it acts. Every action is typed (read, write, send, spend, delete) and checked against a policy you can read at a glance: reads run, writes ask you, deletes are blocked before the request is sent. It connects to your apps over MCP, and `lugh plan` turns a sentence into a program you approve. Rust core, 100+ built-in functions, open source, and you can try it in your browser.





Free
Launch Team

AWS Activate CreditsCredits that fuel your startup’s next milestone
Promoted
Hi Product Hunt! I'm Ash, and I built Lugh.
AI agents can already write good code. What stops most of us from handing them real work is trust: what can this thing touch, and will it check with me before it changes something? Today the answer lives in prompts and config files. In Lugh it's part of the language.
Every action is tagged (read, write, send, spend, delete, submit, execute), and a short policy at the top of each program says what runs on its own, what asks you first, and what's never allowed. Give it a goal in plain English with lugh plan, and an AI writes the program, shows you everything it can touch, and pauses before anything changes. It connects to your apps over MCP, and deletes are blocked before the request ever leaves your machine.
It's open source, the core is Rust, and you can try it in your browser with nothing to install: https://ashutosh3142857.github.io/lugh-lang/
I'd love to hear from you: what would you trust an agent to do if it always asked first?