# AI Context Files (llms.txt)

Krayin publishes two AI context files at the root of this documentation site, following the llmstxt.org (opens new window) convention. Drop them into your project, point your AI coding assistant at them, and the assistant will produce code that follows Krayin's conventions out of the box — no more pasting architecture explanations into every prompt.

The files are companions to the Agent Skills bundle. Agent Skills are richer (full skill files designed to load on demand in Claude Code / Cursor); the llms.txt files are simpler — plain-text context that any AI tool can read, whether or not it supports the skills protocol.

# 🧩 What you get

File URL Use it when...
llms.txt devdocs.krayincrm.com/llms.txt (opens new window) You want a curated index of Krayin's documentation pages, suitable for AI tools that auto-fetch linked content.
llms-full.txt devdocs.krayincrm.com/llms-full.txt (opens new window) You want the complete architectural reference inline — tech stack, package anatomy, Concord model wiring, repository pattern, routes, events, custom attributes, REST API. Drop it directly into the AI's context.

Start with llms.txt. Reach for llms-full.txt once you're doing serious package work or have an assistant with a large context window.

# 🛠️ Supported tools

Any AI coding assistant that can read project files or accept external context works:

Tool How it picks the file up
GitHub Copilot Auto-detects llms.txt / llms-full.txt in the project root.
Cursor Reference it explicitly: @Docs llms.txt in chat.
Claude Code Read it on demand, or pin it into the workspace context.
ChatGPT / Claude.ai Paste the URL or upload the file.
Custom integrations / RAG Ingest the file like any other Markdown source.

For deeper Krayin-awareness in Claude Code or Cursor specifically, also install the Agent Skills bundle — the skills file system is richer than a single text blob.

# ⌨️ 1. Add the files to your project

From the Krayin project root:

curl -O https://devdocs.krayincrm.com/llms.txt
curl -O https://devdocs.krayincrm.com/llms-full.txt

Your project layout becomes:

your-krayin-project/
├── llms.txt
├── llms-full.txt
├── app/
├── bootstrap/
├── packages/
└── ...

That's it for Copilot — it finds the files automatically.

# For a custom package

If you're working primarily inside one of your own packages, drop the files at the package root so the assistant picks them up while you're focused there:

packages/Webkul/Example/
├── llms.txt
├── llms-full.txt
├── composer.json
└── src/
    └── ...

# ⚡ 2. Tool-specific setup

# Cursor

Reference the file explicitly in chat:

@Docs llms.txt

Cursor reads the linked file and uses it as context for the rest of the conversation. See the Cursor docs (opens new window) for the full @Docs directive.

# GitHub Copilot

No setup required — once the file is in your project root or package root, Copilot auto-detects it and uses it as additional context for completions.

# ChatGPT / Claude.ai (web UI)

Either paste the canonical URL into the conversation and ask the assistant to read it, or download the file and attach it as a file upload.

# Claude Code (CLI / IDE extension)

Drop the file in the project root and reference it at the start of your conversation:

Read llms-full.txt and use it as architectural context for this session.

For a richer, on-demand context, prefer Agent Skills — the skills system lazily loads only the bits relevant to the current task.

# 🧪 3. Verify

Open your AI assistant inside the Krayin project and ask it for something that should reflect Krayin conventions:

Scaffold a new package called Webkul/Example with a service provider,
a Concord-registered model, and a repository.

The assistant should:

  • Place files under packages/Webkul/Example/src/.
  • Generate a three-part Concord model (Contract + Model + Proxy).
  • Generate a repository extending Webkul\Core\Eloquent\Repository with model() returning the contract FQCN.
  • Register the service provider in bootstrap/providers.php and the ModuleServiceProvider in config/concord.php.

If the output looks like a generic Laravel package — flat app/ classes, no Contract/Proxy, no Concord registration — the assistant isn't reading the context file. Re-check the file lives in the project root and is named exactly llms.txt / llms-full.txt.

# 🔄 Keep them fresh

Krayin's conventions evolve. Re-download both files when you upgrade Krayin to a new minor version:

curl -O https://devdocs.krayincrm.com/llms.txt
curl -O https://devdocs.krayincrm.com/llms-full.txt

You can commit them to your repo so every contributor's AI tool shares the same context, or .gitignore them and pull on demand — either works.

# 💡 Best practices

  • Pair llms.txt with Agent Skills. Skills give Claude Code / Cursor on-demand, deeply-scoped context; llms.txt covers everything else (Copilot, web ChatGPT, in-house RAG).
  • Customise for heavy modifications. If your project has heavily customised Krayin (forked packages, custom layouts), append a project-specific section to llms-full.txt documenting those changes.
  • Commit it for team consistency. Checked-in context files make sure every developer's AI tool produces the same Krayin-idiomatic suggestions.

# 📝 Next steps

← Architecture concepts →