AI & LLM Integration

Enable AI agents and large language models to access and understand the BrainUs API documentation.

Overview

The BrainUs Developer Portal provides dedicated endpoints for AI agents to:

• Access full documentation in a single text file
• Retrieve individual pages as Markdown/MDX
• Parse structured content for better comprehension

This makes it easier for AI assistants like ChatGPT, Claude, and custom AI applications to help developers integrate with BrainUs API.

Available Endpoints

Full Documentation (/llms-full.txt)

Returns all documentation pages in a single plain text file, optimized for AI consumption.

curl https://developers.brainus.lk/llms-full.txt

Use Cases:

• Training custom AI models on BrainUs documentation
• Providing full context to AI assistants
• Building AI-powered documentation search
• Creating chatbots that answer BrainUs API questions

Individual Pages (*.mdx)

Append .mdx to any documentation page URL to get its content as Markdown.

# Get the Quick Start guide as Markdown
curl https://developers.brainus.lk/docs/quick-start.mdx

# Get Query API reference as Markdown
curl https://developers.brainus.lk/docs/api-reference/query.mdx

# Get error documentation as Markdown
curl https://developers.brainus.lk/docs/errors.mdx

Use Cases:

• Retrieving specific documentation sections
• Building custom documentation viewers
• Integrating docs into AI workflows
• Creating automated documentation tools

Content Negotiation

AI agents can use the Accept header to request Markdown content instead of HTML:

curl -H "Accept: text/markdown" https://developers.brainus.lk/docs/quick-start

When the server detects a preference for Markdown (via the Accept header), it automatically returns the Markdown version.

Implementation Details

The AI/LLM integration uses Fumadocs's built-in features:

1. Processed Markdown

All documentation pages are pre-processed and stored with their Markdown content, enabling fast access for AI agents.

2. Route Handlers

• /llms-full.txt - Route handler that aggregates all pages
• /llms.mdx/[[...slug]] - Route handler for individual pages
• Rewrites - Automatic URL rewriting for .mdx extension

3. Content Format

Each page includes:

• Page title with URL reference
• Full markdown content with proper formatting
• Code examples with syntax highlighting markers
• Structured sections for easy parsing

Example output format:

# Quick Start (https://developers.brainus.lk/docs/quick-start)

Get started with BrainUs API in under 5 minutes...

## Installation

\`\`\`bash
pip install brainus-ai
\`\`\`

...

Building AI Applications

Chatbot Integration

Use the full documentation endpoint to train or provide context to chatbots:

import requests

# Fetch full documentation
response = requests.get("https://developers.brainus.lk/llms-full.txt")
docs_content = response.text

# Provide to your AI model
# (Example with OpenAI)
messages = [
    {"role": "system", "content": f"You are a helpful assistant for the BrainUs API. Use this documentation: {docs_content}"},
    {"role": "user", "content": "How do I query the BrainUs API?"}
]

Documentation Search

Build semantic search over documentation:

from sentence_transformers import SentenceTransformer
import requests

# Load documentation
docs = requests.get("https://developers.brainus.lk/llms-full.txt").text

# Split into sections
sections = docs.split('\n\n')

# Create embeddings
model = SentenceTransformer('all-MiniLM-L6-v2')
embeddings = model.encode(sections)

# Use for semantic search
# ...

Custom AI Tools

Integrate with AI frameworks like LangChain:

from langchain.document_loaders import TextLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.embeddings import OpenAIEmbeddings
from langchain.vectorstores import FAISS

# Load BrainUs documentation
loader = TextLoader("llms-full.txt")
documents = loader.load()

# Split into chunks
text_splitter = RecursiveCharacterTextSplitter(
    chunk_size=1000,
    chunk_overlap=200
)
chunks = text_splitter.split_documents(documents)

# Create vector store
embeddings = OpenAIEmbeddings()
vectorstore = FAISS.from_documents(chunks, embeddings)

# Use for Q&A
# ...

Best Practices

For AI Developers

1. Cache documentation - The /llms-full.txt endpoint is cached and rarely changes
2. Use specific pages - For focused tasks, use individual .mdx pages
3. Include context - Always provide the full page context when using individual pages
4. Respect rate limits - These endpoints are rate-limited like the rest of the site

For AI Agents

1. Parse structure - Use the heading hierarchy to understand document structure
2. Follow links - Internal links point to related documentation
3. Extract code examples - Code blocks are marked with language tags
4. Handle citations - API responses include citations to original sources

Rate Limiting

AI/LLM endpoints follow the same rate limiting rules as the main site:

• Anonymous requests: 100 requests/hour
• Authenticated requests: 1,000 requests/hour (contact us for API keys)

For high-volume AI applications, please contact developers@brainus.lk to discuss dedicated access.

Example Use Cases

1. AI-Powered Support Bot

Build a chatbot that answers BrainUs API questions by loading the full documentation:

import { BrainusAI } from "@brainus/ai";

// Fetch docs once on startup
const docs = await fetch("https://developers.brainus.lk/llms-full.txt");
const docsText = await docs.text();

// Use in chatbot responses
async function answerQuestion(question: string) {
  const context = `Documentation:\n${docsText}\n\nQuestion: ${question}`;
  // Send to your AI model...
}

2. Documentation Assistant

Create a CLI tool that helps developers find relevant documentation:

$ brainus-docs search "how to filter by grade"
Found in: /docs/api-reference/query/filters

Use the QueryFilters object to filter by grade:
...

3. IDE Integration

Build an IDE extension that provides inline documentation:

// Fetch specific page when user hovers over API call
const docUrl = "https://developers.brainus.lk/docs/api-reference/query.mdx";
const markdown = await fetch(docUrl).then((r) => r.text());
// Display in IDE tooltip

Related Resources

• Quick Start - Get started with BrainUs API
• API Reference - Complete API documentation
• Examples - Code examples and integrations