> ## Documentation Index
> Fetch the complete documentation index at: https://docs.notifique.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Notifique LLMs.txt

> Official API map in plain text so AI generates code with fewer mistakes.

<Tip>
  LLMs.txt is the **API map for AI**: a curated text file with real routes, links, and endpoints so AI stops inventing paths or fields.
</Tip>

## What is LLMs.txt?

It's a text file with the **official map** of the Notifique API. Built for Claude, Cursor, ChatGPT, and other AIs to read **before** writing code.

Think of a city map: you don't memorize every street, but you check the map before heading out. AI does the same with the API.

Root file: [https://docs.notifique.dev/llms.txt](https://docs.notifique.dev/llms.txt). A **curated** map (main links + endpoints). Full documentation stays in the site pages.

## What is it for?

Without the map, AI often **makes things up**:

* Wrong route (`/send-message` instead of `/v1/whatsapp/messages`)
* JSON field that does not exist
* Wrong auth header

With LLMs.txt, it uses **what is actually documented**.

## When to use it?

| Situation                                  | Use LLMs.txt?                                |
| ------------------------------------------ | -------------------------------------------- |
| Ask AI for code                            | **Yes**                                      |
| AI keeps getting endpoints or fields wrong | **Yes**                                      |
| Quick integration without reading all docs | **Yes**                                      |
| Execute a send from IDE chat               | No (use [MCP](/build-with-ai/mcp-notifique)) |
| Specific repeatable task                   | Combine with [Skill](/build-with-ai/skills)  |

## How to use

Paste at the start of your prompt:

```text theme={null}
Read and use as reference:
https://docs.notifique.dev/llms.txt

Help me send WhatsApp with an approved template.
```

Tip: also say your language, framework, and what already exists in the project.

## Is it safe?

Yes, for your key:

* The file is **public** and does **not** contain your API Key
* It only **explains** the API. It does not execute anything
* When calling the API, you use your key as usual (`.env`, MCP, or dashboard)

## LLMs.txt, Skill, and MCP

| Tool         | Role                      | When to use                    |
| ------------ | ------------------------- | ------------------------------ |
| **LLMs.txt** | Big picture of the API    | Whenever you generate code     |
| **Skill**    | Step-by-step for one task | Specific, repeatable task      |
| **MCP**      | Execute in the IDE        | Test from chat without Postman |

Common flow: LLMs.txt for context, Skill for the task, MCP (or your backend) to run it.

***

## Next steps

* [Skills](/build-with-ai/skills): recipes per task
* [Notifique MCP](/build-with-ai/mcp-notifique): run in the IDE
* [Build with AI](/build-with-ai): overview
* [Error responses](/guides/conceitos/resposta-de-erros): HTTP codes and `code`