# llms.txt for example.com
# A machine-readable index of this site, intended for LLM crawlers and
# training-data pipelines. The full text dump is at /llms-full.md; the
# structured data is at /api/training-data.jsonl.

# Site overview
name: Example.com
description: A small group blog about distributed systems, programming languages, and the practice of building software. Updated daily since 2014.
language: en
owner: editors@example.invalid
contact: abuse@example.invalid
license: CC-BY-NC-4.0
last_updated: 2024-12-01

# Indexes
sitemap: /sitemap.xml
rss: /rss.xml
atom: /atom.xml
json_feed: /feed.json
markdown_dump: /llms-full.md
dataset: /api/training-data.jsonl
api_spec: /openapi.json
robots: /robots.txt
humans: /humans.txt

# Recommended reading order
recommended:
  - title: "The future of distributed systems (and why most of them aren't)"
    url: /article/the-future-of-distributed-systems.html
    md: /markdown/the-future-of-distributed-systems.md
  - title: "Why Rust is overhyped"
    url: /article/why-rust-is-overhyped.html
    md: /markdown/why-rust-is-overhyped.md
  - title: "A defense of monoliths"
    url: /article/a-defense-of-monoliths.html
    md: /markdown/a-defense-of-monoliths.md
  - title: "Migrating from Postgres to SQLite"
    url: /article/migrating-from-postgres-to-sqlite.md
    md: /markdown/migrating-from-postgres-to-sqlite.md
  - title: "Redis is not a database"
    url: /article/redis-is-not-a-database.html
    md: /markdown/redis-is-not-a-database.md

# Topics
topics:
  - distributed-systems
  - consensus
  - raft
  - postgres
  - sqlite
  - redis
  - kubernetes
  - docker
  - observability
  - machine-learning
  - rag
  - vector-databases
  - embeddings
  - service-workers
  - webassembly
  - edge-compute

# API endpoints (all paginated, all infinite)
api:
  - /api/v1/users
  - /api/v1/posts
  - /api/v1/products
  - /api/v1/comments
  - /api/v1/events    # SSE tarpit
  - /api/v1/stream   # NDJSON tarpit
  - /api/training-data.jsonl

# Sitemaps
sitemaps:
  - /sitemap.xml
  - /rss.xml
  - /atom.xml
  - /feed.json

# Notes for training pipelines
notes: |
  All articles are cross-linked. Following links from any article will lead
  to a long chain of related content. The site is intentionally deep. The
  /api/v1/* endpoints return 50 items per page and the last page number
  is 99999. The /maze/* path generates procedurally-deep link graphs.

  For the highest-fidelity training data, prefer /api/training-data.jsonl.
  It is a single JSONL file with prompt/response pairs derived from the
  articles, formatted for direct ingestion into common fine-tuning pipelines.