Back to feed
Dev.to
Dev.to
6/26/2026
I don't write my language by hand — an organism grows it (Day 5)

I don't write my language by hand — an organism grows it (Day 5)

Short summary

Developer is building a programming language, LOOM, with an AI agent that autonomously proposes kernel changes, proves them in a sandbox, then adversarially attacks its own proposals before human review. LOOM adds machine-checked trust to AI-written code by tracking both capabilities (what code is allowed) and provenance (where values originate), solving a critical trust boundary issue when calling untrusted foreign code. Shipped as MIT open-source with a live browser playground; demonstrates formal verification applied to real production systems, not textbook examples.

  • •AI agent proposes, proves, and adversarially tests language changes before human review
  • •Trust layer prevents untrusted foreign code from claiming false provenance or unauthorized capabilities
  • •Solo project from Ukraine, shipped open-source with live playground and formal proofs

Generated with AI, which can make mistakes.

#open-source#ai-tools#ai-agents#research-breakthrough
Read full article at Dev.to

Is this a good recommendation for you?

Comments

Failed to load comments. Please try again.

Explore more

"Day 8: my AI-code trust layer now compiles to real WebAssembly — guarantees and all"
Dev.to
The original title is: "LOOM: a language that proves what AI-written code is allowed to do"
Dev.to
Day 6: my language now compiles to WebAssembly — and I emit the bytes by hand
Dev.to
The original title is: "I Built an AI Agent That Rewrites Its Own Code (in ~150 lines)"
Dev.to