llms.txt and llms-full.txt

Two plain-Markdown files, regenerated on every deploy, designed to be handed straight to a language model.

• https://gabegiro.com/llms.txt, a curated index following the llmstxt.org convention. Links and one-line summaries. Small enough to paste anywhere.
• https://gabegiro.com/llms-full.txt, the full knowledge file. Everything in the index plus prices, FAQ answers and descriptions inline, so a model does not need to follow a single link.

Which one to use

If you are configuring a chat assistant, a support bot or a retrieval index against this site, use llms-full.txt. It is self-contained.

Use llms.txt when you want a table of contents rather than the contents: agent tooling that will fetch pages itself, or a human orienting quickly.

Structure of llms.txt

# Gabe Giro - AI, Mobile & Full-Stack Engineer

> one-line summary

intro paragraph

## Services
- [Service name](url): description - price summary

## Documentation
- [Doc page title](url): description

## Case Studies
- [Title](url): description

## Articles
- [Title](url): description

## Key Pages
- [Label](url): meta description

---

contact line

Sections are omitted entirely when empty rather than rendered as an empty heading.

Structure of llms-full.txt

Same header, then the same material expanded. Each service becomes a ### block carrying its description, canonical URL, a Pricing: list of every tier, and that page's FAQ as question and answer pairs. Case studies, documentation pages and articles each get a block with description and URL. A general FAQ section at the end carries the questions that are not tied to a specific service.

How they are generated

At the end of every build, a small integration walks the built HTML, extracts the JSON-LD each page already emits and the page titles and meta descriptions, and assembles both files from that. There is no separate content source, which is the point: the files cannot describe a service that no longer exists, or quote a price the page does not show.

The only hand-written parts are the four curated header fields, the title, the one-line summary, the intro paragraph and the contact line. Everything else is derived.

Guarantees and caveats

Guaranteed: the files exist at those URLs on every deploy, are valid UTF-8 Markdown, and never contain a promotional price. Prices are always the standing list price.

Not guaranteed: section ordering within a category is by name or date and may shift as content is added. Do not parse these by line number. If you need something stable to key off, the URLs are it.

Caveat worth knowing: these files are built artifacts. They exist after a production build, not in a local development server. If you clone this site and run it locally, they will not be there, and nothing is broken.