STIG-ØRJAN SMELROR / BENALMÁDENA, SPAIN

Between the model and the machine.

The demo starts at the model. The work starts underneath it: context, boundaries, state, authority and everything a real machine can do after the prompt.

ORIGIN
Northern Norway
BASE
Andalucía
NATIVE ENVIRONMENT
Linux

SELECTED WORK

Tools for the layer below the prompt.

Different projects, one recurring concern: give an agent structure before action, and give the person responsible something they can inspect.

AGENT ENVIRONMENT / CONTEXT

Zaguán Blade

A local-first coding environment built around context that survives.

Files on disk as truth, project memory, symbol indexing and long-running conversations.

RUST / TAURI / REACT

WRITING

Notes from the work.

Ideas about systems, AI, software and what happens when they meet the real world.

GPT-6.1 Terra: a fast model in a crowded roomGPT-6.1 Terra landed on OpenAI's deployment safety hub on 26 October 2026 as an addendum to the GPT-6 Astra system card (https://deploymentsafety.openai.com/gpt-6-1-terra). It is the mid-tier model in the GPT-6 series. It sits below the Sol and Luna variants in capability, and above nothing, since the slot has been empty. I think it handles medium and small tasks "pretty decently" for my needs.↗When the Model Writes Your Own ExploitsI have built and run production systems for over thirty years. The part that keeps me up at night has moved. It used to be configuration drift, library rot, operator error. Those are all still there. But now a model can find a chain of five bugs across a browser sandbox, a kernel, and a payment library before you have finished your coffee.↗
All writing →

A LITTLE CONTEXT

Trust, but verify.

I am Stig-Ørjan Smelror, a Norwegian systems builder living in Spain. After more than thirty years around computers, I still want to know how the machinery works underneath.

New is not automatically better. Old is not automatically wise. Systems earn their place by working.

Find me on X ↗