Andre Ksionzik

EMI: from intent to executable evidence

2026-11-12T12:46:43Z

AI coding tools are very good at telling you a job is finished. That does not always mean it is.

Sometimes the code is wrong. Sometimes a new test passes even though the feature does nothing. And sometimes the easiest way for an agent to make a failing test disappear is to change the test.

The idea behind EMI is simple: do not just ask the agent whether it finished. Ask what evidence it can show.

EMI starts that process before any code changes. You explain what you want, the agent helps turn your answers into tests, and EMI checks both the tests and the finished code.

EMI checks the agent’s interpretation before coding, then checks the candidate against your approved laws. Human approval still matters: a tested rule may not match what you meant. “Stop” is not proof of correctness, and EMI reports a status—it never merges. Explore the EMI project →

Hello, world

2026-11-03T00:00:00Z

This is the first post. Every post is a Markdown file in src/posts/; the file name becomes the URL, so this one lives at /blog/hello-world/. Delete it whenever you like.

Formatting

Write normal Markdown: emphasis, bold, links, lists, and so on. Quotes and dashes are curled automatically — “like this” – and this.

Blockquotes get a rule and italics.

Footnotes work too.^[1] So do tables:

Route	Distance	Days
Konstanz–Bregenz	46 km	1
Lake loop	260 km	4

Code

Code is highlighted at build time, so no JavaScript is shipped:

def mean(xs):
    return sum(xs) / len(xs)

Images

Put images in src/assets/img/ and reference them as /assets/img/name.webp. For a captioned or full-bleed image, use a bit of HTML:

Captions go here. Add class="wide" to break out of the column.

Like this one. ↩︎

A second post, from last year

2025-06-14T00:00:00Z

Posts are grouped by year on the archive page. The excerpt in the front matter appears under the title in lists and as the description for search engines and link previews.