Agentic Web Ready

Powering the Agentic Web

QRTRAC is built for both humans and AI agents. We provide structured, machine-readable data to ensure accurate retrieval and seamless automation.

llms.txt

A compressed markdown index following the llms-txt.org standard. Perfect for crawlers.

Read Standard

llms.json

Full structured dataset with semantic chunking optimized for RAG systems and vector DBs.

Download Manifest

llms-full.txt

The complete QRTRAC context in a single flattened text file for large window LLMs.

View Context

/llms/ Directory

Direct URL-matched raw Markdown generation of our entire application site mapping.

Explore Directory

Implementation Patterns

For Retrieval (RAG)

Point your scraping agents to /llms.json. Each entry is pre-chunked to ~1200 characters to fit perfectly into context windows while preserving semantic context.

For Discovery

Search agents should use /llms.txt to prioritize crawling. We've included priority descriptions to help agents find pricing and API docs first.

For Deep Integration

Workflows requiring specific page context can map their requests dynamically via the /llms/ mirror to ingest native, noise-free Markdown.

Ingestion Best Practices

How you ingest QRTRAC documentation depends on your agent's architecture. Here is a breakdown of the recommended approach for each pattern:

Use Case Recommended Resource Notes
RAG / Vector Store Indexing /llms.json Each entry is pre-chunked to approximately 1,200 characters with a url, title, and content field. Embed the content field directly — no additional splitting required.
Single Large-Context LLM Pass /llms-full.txt The complete site context in one flat Markdown file. Best for models with 100k+ token context windows (Claude, GPT-4o, Gemini 1.5). Append your task instructions after the file contents.
Priority-Aware Crawling /llms.txt Follows the llms-txt.org standard. Each section is labeled with priority tags. Pricing, API documentation, and getting started guides are listed first. Use this as your crawl queue seed.
Per-Page Dynamic Retrieval /llms/{path}/index.md Every public page has a matching Markdown mirror at its URL path. Agents can dynamically fetch clean, noise-free content for any page by substituting the URL path into the /llms/ prefix.

Crawler & Bot Instructions

Configure your automated agents with the correct User-Agent strings and ingestion rules for QRTRAC resources.

Permitted Crawlers

All well-behaved crawlers following standard robots.txt rules are permitted to index QRTRAC public pages. AI training crawlers are governed by our robots.txt file. Agents that respect Crawl-delay headers are preferred.

User-agent: *

Allow: /llms/

Allow: /llms.txt

Allow: /llms.json

Allow: /llms-full.txt

Recommended Request Headers

When fetching /llms.json programmatically, send an Accept header and identify your agent via User-Agent. Rate limit your requests to a maximum of 10 requests per second to avoid triggering DDoS protection.

GET /llms.json HTTP/1.1

Host: qrtrac.com

Accept: application/json

User-Agent: MyRAGBot/1.0

API & RAG Schema Reference

The /llms.json manifest follows a predictable schema for semantic indexing. Each document entry exposes the following fields:

Document Entry Schema

"url": "https://qrtrac.com/...",

"title": "Page title string",

"description": "Meta description",

"content": "Clean Markdown...",

"lastModified": "ISO 8601 date",

"section": "help | solutions | ..."

Semantic Search Indexing Tips

  • → Embed the content field for semantic similarity search. Concatenate title + description for title-boosted retrieval.
  • → Store url as metadata — return it in citations so users can navigate to the source page directly.
  • → Filter by section to restrict retrieval to help documentation, solutions pages, or use-case articles based on user query intent.
  • → Refresh the index by re-fetching /llms.json weekly or whenever lastModified changes.