Reflex

Appearance

AI-Assisted Coding with llms.txt ​

Get better AI-generated Reflex code by feeding your assistant the right context.

Reflex provides an llms.txt file — a concise reference that teaches AI coding assistants (Cursor, Copilot, Claude, ChatGPT, etc.) the library's architecture, conventions, and patterns. Using it dramatically improves the quality of generated code.

What is llms.txt? ​

llms.txt is a single-file summary of Reflex's best practices designed specifically for LLM consumption. It covers:

  • State architecture — recommended file structure and naming
  • Event conventions — synchronous handlers, effect tuples, current() usage
  • Subscription patterns — root vs. computed subs, view-ready data
  • Component contract — minimal subscriptions, dispatch-only user intent
  • Effect/coeffect rules — isolating I/O
  • Testing patterns — event and subscription test examples
  • Starter skeleton — a complete copy-paste pattern

How to Use ​

Create a project-level rule that automatically includes the Reflex conventions in every AI interaction:

  1. Create .cursor/rules/reflex.mdc in your project root:

    markdown
    ---
description: Reflex state management conventions
globs: "**/*.{ts,tsx,js,jsx}"
---

@https://raw.githubusercontent.com/flexsurfer/reflex/main/llms.txt

  2. The AI assistant will now follow Reflex patterns whenever you work on matching files.

Option 2: Paste into Chat Context ​

Copy the contents of the file and paste it directly into your AI assistant's chat:

  1. Open llms.txt
  2. Copy the entire contents
  3. Paste it at the beginning of your conversation (or attach as a file)
  4. Then ask your question or describe the feature you want to build

Option 3: Reference the URL Directly ​

Many AI tools support fetching URLs. You can reference the file directly:

  • Cursor: Use @https://raw.githubusercontent.com/flexsurfer/reflex/main/llms.txt in chat
  • Claude: Attach the URL or paste the contents
  • ChatGPT: Paste the contents or provide the URL for web-browsing-enabled models

Option 4: Add to Your Project ​

Download the file into your repository so it's always available as local context:

bash
curl -o llms.txt https://raw.githubusercontent.com/flexsurfer/reflex/main/llms.txt

Then reference it from your AI tool (e.g., @llms.txt in Cursor).

Example Prompts ​

Once your AI assistant has the llms.txt context, try prompts like:

"Create a new notifications feature with events for fetching from API, marking as read, and a subscription for unread count."

"Add a settings slice to the app DB with events for updating theme and language, persisted to localStorage via effects."

"Write tests for my todos/add event handler and todos/open_count subscription."

The AI will follow the correct patterns: namespaced event IDs, effect tuples for side effects, current() for draft data, view-ready subscriptions, and proper component contracts.

Keeping Up to Date ​

The llms.txt file is maintained alongside the Reflex library. When referencing it via URL, you always get the latest version. If you've downloaded a local copy, update it periodically:

bash
curl -o llms.txt https://raw.githubusercontent.com/flexsurfer/reflex/main/llms.txt