Short, factual write-ups of things I have built with AI agents. Each note explains how the work was split between me and the agents, which rules and checks applied, what went wrong and what the result does not prove. My client work as a Forward Deployed Engineer stays private; these are projects I can show.
A read-only portfolio tracker built across seven repositories, where Codex implements under written rules, a public Docusaurus portal documents the system and by rule, every substantial change goes from a spec to a version decision.
How a Spanish-language course on OKR, Scrum and DevOps moved from ChatGPT and Google Drive to a repository with eight agent roles, and what those agents produced under rules that keep the approvals with me.
How one dependency-free script generates the bilingual profile, and how written rules, two hooks and build checks keep AI agents from editing the wrong file or claiming the wrong thing.
Static site
Agent workflow
Verification
Consent
More notes may follow as other projects are written up.
Your privacy choices
With your permission, optional tools help us understand visits and improve this site. You can choose each purpose or continue without them.