MCP & llms.txt

Configure your AI agent to access the Appmax documentation — via plain text file or MCP server with 13 integrated tools.

llms.txt

We provide the entire Appmax API documentation in a single plain text file, following the llms.txt standard, to make it easy for language models (LLMs), AI agents, and automation tools to consume.

Download

llms.txt (English)

Full documentation in plain text (~262 KB, 54 sections).

llms.txt (Portugues)

Complete documentation in Portuguese plain text (~269 KB, 54 sections).

What is llms.txt?

llms.txt is an open standard that allows websites to provide their content in a format optimized for LLMs. Similar to robots.txt for crawlers, llms.txt makes it easy for AI agents to access and understand documentation efficiently.

What's included

The file contains all documentation pages in clean Markdown format:

• Guides — Quickstart, getting started, authentication, installation & callback, external-id, webhooks, rate limit, environments, publishing to production, FAQ, AI & MCP
• Examples — Full integration, installments, recurring, checkout
• API Reference — Customers, orders, payments (credit card, Pix, boleto, Apple Pay, Apple Pay merchant session), refunds, subscriptions, payment links
• Payment split — Guides (overview, status, approved banks, FAQ) + endpoints (recipient, facematch, order split, balances, anticipation, withdrawal)

How to use

With Claude, ChatGPT, or another LLM

Paste the file content as context in your conversation:

Here is the Appmax API documentation:

[llms.txt content]

Based on this documentation, help me implement...

With AI agents

Point your agent directly to the URL:

https://docs.appmax.com.br/llms-en.txt

MCP server

Open standard

Our server implements the Model Context Protocol — the open standard maintained by Anthropic for connecting AI agents to external data sources and tools. Any MCP-compatible client (Claude Code, Claude Desktop, Cursor, Windsurf, VS Code with Copilot, among others) connects without any proprietary adapter.

The server exposes 13 tools:

Documentation

Tool	Description
list_pages	List available pages. Supports prefix filter (e.g., api- to list only endpoints).
get_page	Get the full content of a page by ID
search_docs	Search the documentation by term (top 10 by relevance)
get_full_docs	Get the entire documentation as plain text (~140 KB)
check_health	Server status and documentation stats

Diagnostics

Tool	Description
diagnose_error	Diagnose HTTP errors (401/422/429/500) with likely root cause and suggested fix
validate_payload	Validate a JSON payload against the endpoint schema before sending to the API
validate_order_total	Verify order total calculation locally
validate_installation_flow	Audits the installation flow implementation (4 steps + validation URL) from project snippets, reporting pass/fail per step

Code generation

Tool	Description
generate_code_snippet	Generate a ready-to-use snippet in curl/Node/Python/PHP/Go for any endpoint
get_integration_flow	Get a step-by-step flow (checkout, installation, subscription, etc.)

Onboarding and webhooks

Tool	Description
get_onboarding_checklist	Personalized admin checklist by integration type and stage
get_webhook_schema	Typed schema + example payload for any of the 28 webhook events

Health check

Before using the MCP, verify the server is operational:

curl https://m7nwi1m199.execute-api.us-east-1.amazonaws.com/health

Expected response:

{
  "status": "healthy",
  "name": "appmax-docs",
  "version": "1.0.0",
  "tools": ["list_pages", "get_page", "search_docs", "get_full_docs", "check_health",
            "diagnose_error", "validate_payload", "validate_order_total",
            "validate_installation_flow", "generate_code_snippet",
            "get_integration_flow", "get_onboarding_checklist", "get_webhook_schema"]
}

Configuration

Claude Code

Add to .mcp.json in your project root:

{
  "mcpServers": {
    "appmax-docs": {
      "type": "http",
      "url": "https://m7nwi1m199.execute-api.us-east-1.amazonaws.com/mcp"
    }
  }
}

Claude Desktop

Claude Desktop still connects natively over stdio, so you need the mcp-remote shim (installed on demand via npx) to bridge stdio ↔ HTTP.

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "appmax-docs": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://m7nwi1m199.execute-api.us-east-1.amazonaws.com/mcp"
      ]
    }
  }
}

Requires Node.js installed on your system (npx ships with it).

Any MCP-compatible client

Point to the endpoint:

https://m7nwi1m199.execute-api.us-east-1.amazonaws.com/mcp

Quick test

Once configured, test with a simple call. Example with search_docs:

You: "How does the payment confirmation webhook work?"

The agent calls: search_docs({ "query": "webhook payment confirmation" })

Returns: relevant excerpts from the webhooks and order status guides.

If the agent answered with Appmax documentation content, the connection is working. You now have access to all 13 tools.

See the full reference for all tools →

Next steps

Why use AI?

See the benefits, demo videos, and how AI accelerates your integration.

Quickstart

Prefer to write code by hand? Jump straight into the quickstart.