How an AI Sees the Antikythera Project Code
At the .IA Antikythera Island Project, we talk a lot about the future of human-machine collaboration. But are we consistent with that in our own technological guts?
Recently, we submitted our public repository to a peculiar audit: we asked Google Code Wiki (an experimental AI tool) to read our code and explain what this project is about.
The result was fascinating, not only because of what it got right, but because of what this means for the transparency of our digital "Country Code".
What the AI "Saw" #
Without giving it any prior context, the AI analyzed our files and correctly deduced:
- The Technical Architecture: It immediately identified that we use Eleventy (11ty), a static site generator, prioritizing speed and simplicity over heavy frameworks.
- The Dual Mission: It grasped that we are not just a technical website, but an initiative with a historical and activist component. It understood that the code serves a narrative: the connection between the Antikythera mechanism and modern AI.
- The Content Structure: It "read" our folder organization and understood how we separate the logic from the blog post publication.
Why Does This Matter? #
If an AI can understand your project just by reading the code, it means your structure is logical, clean, and honest. It is the ultimate "Semantic SEO" test for the era of autonomous agents.
As a result of this experiment, we have added a CONTRIBUTING.md file (which the AI noted was missing) to make it easier for both humans and future agents to collaborate with us more fluidly.
Conclusion #
We hide nothing. Our code is open not just by license, but by readability. We want that, in 2000 years, if someone (or something) finds this digital repository "shipwrecked" on the net, they can understand it as well as we understand the gears of Antikythera today.
You can read our new contribution guide here.
- Previous: I Working Session with Gemini 3.0 Pro