Using AI Tools for Documentation

Last modified: October 8, 2026

Introduction

Contributors may use AI tools such as large language models and writing assistants to draft or improve documentation. However, these tools must be treated as writing aids, like spell-checkers or linters, and not as authoritative sources.

Contributor Responsibility

If you use AI tools, before contributing you must:

  • Review and verify the accuracy of all AI-assisted content.
  • Ensure the content does not infringe third-party copyrights.
  • Remove hallucinated content or unsupported claims.
  • Confirm that the AI tool's terms do not impose restrictions inconsistent with the Mendix Contributor License Agreement.

AI output must be treated as unverified draft material.

Maintainer Discretion

Maintainers may request clarification, edits, or removal of AI-assisted content if concerns arise regarding accuracy, licensing, or quality.

Using AI Assistants

This repository is configured for use with Claude Code, an AI-powered coding assistant. It also includes customization for GitHub Copilot.

Claude Code Configuration

The shared Claude Code settings for this repository are in .claude/settings.json. They define the following:

  • Permissions that allow or deny specific commands and tools
  • Guardrail hooks that check commands before they run
  • A status line
  • Telemetry turned off

Personalizing Claude Code

These settings do not configure or mandate any specific provider or language model. You need to set up your own access to a model, either by signing in to Claude Code or by configuring a provider. To add or override settings, create .claude/settings.local.json in the root of your repo clone. This file overrides the shared settings. It is gitignored, so Git doesn't commit it.

To get started, copy .claude/settings.local.json.example to .claude/settings.local.json and edit the values. The example shows one setup that uses Amazon Bedrock. Delete any keys you don't need. See the Environment variables page of the Claude Code documentation for information on environment variables.

Fixing Truncated Output

If you are updating a lot of documentation, Claude Code may truncate its output. In this case, increase the output token limit by setting CLAUDE_CODE_MAX_OUTPUT_TOKENS to a higher value in your .claude/settings.local.json file. You can find more information about CLAUDE_CODE_MAX_OUTPUT_TOKENS on the Environment variables page of the Claude Code documentation.

Custom Skills

This repository includes custom Claude Code skills optimized for documentation work:

  • /docs-proofread – Checks spelling, grammar, punctuation, and basic Markdown formatting
  • /docs-polish – Applies the Mendix style guide and improves clarity, readability, and word choice without changing meaning
  • /docs-enhance – Performs comprehensive editing including reorganization, restructuring, and stronger phrasing
  • /docs-add – Adds new content to an existing page while preserving original structure
  • /docs-review – Analyzes a page and generates suggestions for improvements without making any edits
  • /docs-pr-review – Reviews all changes in a PR rather than just a single document
  • /docs-alt-text – Suggests W3C-compliant alt text for images on a page

These skills are available to all contributors using Claude Code with this repository. If you use GitHub Copilot, use the prompt files described in the next section instead.

GitHub Copilot Configuration

If you use GitHub Copilot, the repository provides the following:

Read More