# promptito

> An LLM-first blog. Content structured for machine consumption: dense, semantic, parseable. Each post available as Markdown with YAML frontmatter and as JSON-LD (Schema.org TechArticle). Syndicated via JSON Feed 1.1.

- Format: Markdown + YAML frontmatter (source), JSON-LD (metadata), JSON Feed 1.1 (syndication)
- License: CC-BY-4.0
- Feed: [JSON Feed](feed.json)
- Source: [GitHub](https://github.com/cuihtlauac/promptito)

## Format

- [Promptito Post Format Specification](SPEC.md): Self-contained spec for the structured post format. An LLM can generate a valid post from this document alone.

## Authoring

- [Authoring Workflow](CLAUDE.md): Two-step pipeline — ingestion (human provides facts, agent structures into post.md) and export (node export.mjs <lang> <slug> generates natural language from the structured post). CLAUDE.md is the canonical reference for LLM agents writing promptito posts.

## Posts

- [Origin of y-skill: How a Tired Cat Became a Skill That Writes Skills](posts/y-skill-origin/post.md): Origin story of y-skill, a self-referential Claude Skill implementing the Y combinator: a SKILL.md whose only supporting file is itself and whose payload mints further skills. The chain: the &#34;cat sat on the mat&#34; attention example, reframed as oriented bonds between tokens; four context-dependent word classes mapped to lambda-calculus constructs (attention as a probabilistic interpreter); a natural-language Y-combinator sentence mirroring λf.(λx.f(x x))(λx.f(x x)); the sentence rebuilt as a skill; a repository with structural and behavioural gates, an OCaml oracle, Turing-completeness constructions, and a defense of the stochastic interpreter (computers err too; LLMs hallucinate where silicon crashes or hangs). The narrative version (ORIGIN.md) was ghost-written by Claude Code from the two founding chat transcripts.
  - Human-readable version (en): [rendered](https://cuihtlauac.pages.dev/read.html?p=posts/2026-10-07-y-skill-origin/human.en.md), [markdown source on GitHub](https://github.com/cuihtlauac/promptito/blob/main/posts/2026-10-07-y-skill-origin/human.en.md) — not deployed here; lives in the source repository
- [Bootstrapping the Promptito Blog](posts/llm-first-blog/post.md): promptito is a blog designed for LLM consumption, not human reading. Posts are structured as Markdown with YAML frontmatter (authoring), JSON-LD TechArticle (semantic metadata), and JSON Feed 1.1 (syndication). The site exposes llms.txt and llms-full.txt for discovery. Machine readability is paramount: density, structure, semantic clarity.
  - Human-readable version (en): [rendered](https://cuihtlauac.pages.dev/read.html?p=posts/2026-03-25-llm-first-blog/human.en.md), [markdown source on GitHub](https://github.com/cuihtlauac/promptito/blob/main/posts/2026-03-25-llm-first-blog/human.en.md) — not deployed here; lives in the source repository