---
title: Using these docs with AI
description: Three ways to put this documentation in front of an AI assistant — copy it, point it at a URL, or connect it directly.
updated: 2026-07-30
order: 2
related: [/reference/mcp, /reference/api]
---

This documentation is written to be read by assistants as well as people. Every page has a
Markdown twin, and the whole set is small enough to hand over in one go — so you can ask
questions about PostbackX instead of hunting for the page that answers them.

Three ways to do it, from least setup to most useful.

## Copy the whole thing

All twelve pages, about 12,000 tokens. That fits comfortably in any current assistant's
context window, so it can answer questions spanning several pages at once — how attribution
relates to postbacks, say, or what to check when conversions go missing.

Copy the whole documentation set from https://docs.postbackx.com/llms-full.txt

Paste it into a new conversation and ask your question. There is nothing to configure and no
account involved.

## Give it the URL

If your assistant can browse the web, it does not need the paste:

```
https://docs.postbackx.com/llms-full.txt
```

That single file is the entire documentation set as plain text. Ask it to read that, then ask
your question.

## Connect it to your account

The other two options give an assistant the *documentation*. This one also gives it **your
numbers**.

PostbackX runs an [MCP server](/reference/mcp). Once connected, an assistant can read your
campaigns, spend and conversions directly — and it can read this documentation too, since the
docs are exposed as resources on the same connection. No copying, and nothing to keep up to
date.

That combination is what makes the difference: an assistant that knows both what ROAS means
here *and* what yours was last week can answer questions neither source could alone.

## The machine-readable files

| File | What it is | Size |
| --- | --- | --- |
| [`/llms.txt`](/llms.txt) | Index of every page with one-line summaries. For picking what to read. | ~1,000 tokens |
| [`/llms-full.txt`](/llms-full.txt) | Every page, concatenated. For reading in one fetch. | ~12,000 tokens |
| `<page>.md` | Any single page as Markdown — add `.md` to its URL. | ~1,000 tokens |
| [OpenAPI spec](https://api.postbackx.com/v1/openapi.json) | The REST API's endpoints and parameters. Describes the API, not this prose. | — |

Each page's Markdown twin keeps its frontmatter, so a title, description and `updated` date
travel with the text.

## Worth knowing

**A copy is a snapshot.** Documentation you pasted last month describes the product as it was
last month. Every page carries an `updated` date so an assistant can tell you how old its
information is — and re-copying costs one click.

**Never take figures from documentation.** Payouts, spend, ROAS and conversion counts in these
pages are illustrative examples, not your data. An assistant reading only the docs will
happily quote them at you. For real numbers it needs
[a connection to your account](/reference/mcp) or the [API](/reference/api).

**Check the vocabulary.** This industry uses the same words for different things, and an
assistant trained on the wider web tends to say "network" where PostbackX says
[offer source](/concepts/vocabulary). If an answer sounds subtly off, that is usually why.