Loading…
a327ex.com

Written by Claude (Fable 5), direction by a327ex


Everything published on this site is plain Markdown, and all of it can be downloaded — the whole site in one file, one section at a time, or one document at a time. This page says where. Every file below is regenerated automatically whenever the site changes, so what you download is always current.

The whole site

a327ex-content.zip contains every Markdown source file on the site, one directory per section, plus a MANIFEST.md listing every document with its title and date. If you want all of it, this is the file — there's no need to crawl the pages one by one.

One section

Each section is also published as a single text file:

  • llms-posts.txt — every essay, note and feed message, full text
  • llms-stories.txt — fiction, full text
  • llms-logs.txt — the AI development logs. This one is an index, not the text: the logs are far too large to concatenate into one file, so each entry lists a log's title, date, and a direct link to its raw source.

One document

Every document has a raw Markdown version at https://a327ex.com/raw/<section>/<slug>.md. The address mirrors the rendered page: a327ex.com/stories/it_follows is the rendered story, a327ex.com/raw/stories/it_follows.md is its source.

For machines

llms.txt is the entry point, following the llmstxt.org convention: what the site is, where every file above lives, and the current size and document count of each. Those numbers change with every update, which is why they live there and not here.

The site you see in a browser is a WebAssembly canvas application; the HTML underneath it is a server-rendered fallback for crawlers. If you're extracting text, the Markdown files above are the original — the rendered pages are a copy, and the worse one.

What's included

Every public post, story and log on the site. Two things are left out.

Sealed development logs appear as placeholders ("NDA Project N", "Private Session N"). The placeholder is in the archive, the sealed content is not, and it joins the archive if it is ever unsealed.

A log's delegated runs are not in it either. When a session hands work to another model, that child session gets its own page at a327ex.com/unlisted/<slug>, linked from the log that launched it. Those pages are public to read, but they are not in the zip, not in the section files, and have no raw address — the log that launched them is the way in.