# AI-Powered Development with Corex UI
Corex UI provides machine-readable documentation at https://corex-ui.com/llms.txt
so AI coding assistants can deliver accurate component implementations with built-in accessibility and design system patterns.
# What is LLMs.txt?
LLMs.txt is a standardized format optimized for AI consumption. It includes component APIs, usage patterns, accessibility guidelines, and design tokens in a structured format that reduces hallucinations and aligns AI suggestions with Corex UI conventions.
Benefits: Accurate prop suggestions, accessibility-first code, consistent design token usage, and idiomatic patterns.
# Quick Setup
Cursor
@Docs https://corex-ui.com/llms.txt
Windsurf (.windsurfrules
):
docs:
- https://corex-ui.com/llms.txt
Claude Projects
Add the URL contents to Project Knowledge and reference it in conversations.
ChatGPT Custom GPTs
Upload LLMs.txt content to the knowledge base during GPT creation.
GitHub Copilot (.github/copilot-instructions.md
):
Reference Corex UI documentation: https://corex-ui.com/llms.txt
Use design tokens, include accessibility, follow composition patterns.
Continue (~/.continue/config.json
):
{
"contextProviders": [
{
"name": "url",
"params": {
"url": "https://corex-ui.com/llms.txt"
}
}
]
}
# Effective Prompts
Component Implementation
Using Corex UI from https://corex-ui.com/llms.txt,
create a [component] with:
- Proper accessibility (ARIA, keyboard nav)
- Design token styling
- Minimal, idiomatic code
Refactoring
Refactor this to match Corex UI patterns from https://corex-ui.com/llms.txt:
[paste code]
- Use design tokens
- Add accessibility features
- Follow official composition patterns
Debugging
This Corex UI [component] isn't working correctly:
[paste code]
Compare to https://corex-ui.com/llms.txt and fix the issues.
# Best Practices
Always include the LLMs.txt URL in prompts to ground the AI in official documentation. Request minimal examples to avoid over-engineering. Verify that generated code uses design tokens instead of hardcoded values and includes proper accessibility attributes.
For team workflows, create a shared prompt library. For code reviews, check that AI-generated components follow Corex patterns and include keyboard navigation and ARIA attributes.
# Troubleshooting
AI suggests non-existent props: Ask it to re-check the documentation and use only documented APIs.
Missing accessibility: Request explicit ARIA attributes, keyboard handlers, and focus management per the official patterns.
Wrong styling: Ask it to replace all hardcoded styles with design tokens from the documentation.
Cached documentation: Re-add the URL to your tool after Corex UI updates.
# Resources and Community
- LLMs.txt (machine-readable):
https://corex-ui.com/llms.txt
- Documentation (human-readable):
https://corex-ui.com/docs
- Component Examples:
https://corex-ui.com/examples
- GitHub Repository:
https://github.com/corex/ui
- Discussions:
https://github.com/corex/ui/discussions
The LLMs.txt format is continuously improved based on community feedback. If you discover cases where AI assistants generate incorrect code, hallucinate APIs, or miss important patterns, please open an issue with the llm-docs
label. Your feedback helps improve the documentation for everyone.
We also welcome contributions for prompt patterns that work well. If you've developed effective ways to interact with AI assistants using Corex UI, consider sharing them in GitHub Discussions so other developers can benefit from your experience.
The documentation is versioned alongside Corex UI releases. Each update includes improvements to component descriptions, additional usage examples, and refined patterns based on real-world usage. Check the changelog section in LLMs.txt to see what's changed between versions.
For questions about integrating Corex UI with specific AI tools not covered in this guide, or for help troubleshooting integration issues, join the community discussions. The team and community members actively help with setup questions and share configuration tips for various development environments.
When reporting issues with AI-generated code, include the tool you're using, the prompt you provided, and the generated output. This context helps identify whether the issue is with the LLMs.txt content, the AI tool's interpretation, or the prompt structure itself.