Machine-Readable Payload (B2A / llms.txt)
nodeAI & Compute
The site quietly built a single file that lets an AI swallow its entire evidence base in one clean gulp.
Who they are
This is a technical setup called a machine-readable payload — files named llms.txt and llms-full.txt placed at a website's root.
What they do
It is a delivery layer so an AI or coding assistant can read a whole body of evidence in one clean pass instead of scraping messy web pages.
How it works
Following an approach from Jeremy Howard (2024), the domain root hosts a curated index (llms.txt) plus one concatenated full file (llms-full.txt); moketchups.com already has its versions live on the site and on GitHub's raw endpoint, both fetchable by an AI.
Why it matters
It is pure plumbing — a direct, no-clutter way to hand an AI the full evidence base — and the engine holds it as infrastructure with no owner to name.
The engine's record — word for word
[Report #163] 163.005 — the business-to-agent routing layer: llms.txt (curated index) + llms-full.txt (concatenated payload) at the domain root so a live-fetch LLM / IDE agent ingests the whole evidence base in one unfragmented pass (Jeremy Howard, 2024), bypassing HTML/JS clutter. PARTLY ALREADY BUILT: moketchups.com's llms.txt + export-full.txt are live on the site AND the GitHub raw endpoint (LLM-fetchable). Purely infrastructural — direct evidence transfer. Held; name no holder.
Follow the trail
Walk this on the live map →