AI Comprehension Tool

OKF Bundle Generator

Turn important website pages into an organized Open Knowledge Format v0.2 bundle. Create readable Markdown concepts with source links and metadata, review their quality, then download a portable ZIP.

Public websites only. Enter the domain you own or have permission to process.

What is Open Knowledge Format?

Open Knowledge Format, or OKF, is a way of packaging knowledge as Markdown documents with a short YAML metadata block at the top of each concept. The files are organized into folders and can link to one another. This makes the same information readable by people and straightforward for compatible retrieval systems to process.

Instead of putting an entire website into one long file, OKF separates subjects into individual concepts. A service page may become one concept, a product page another, and a detailed explanation a third. An optional index.md gives readers a map of the available knowledge without forcing them to load everything at once.

The original OKF v0.2 specification describes a minimally opinionated format. It does not mandate a central registry, special AI model or proprietary service.

How an OKF knowledge bundle is structured

A concept document begins with frontmatter describing the kind of knowledge it contains, followed by ordinary Markdown. The type field is the only field required for every concept in v0.2. Useful optional fields include title, description, resource and tags.

Example bundle

okf/
  index.md
  concepts/
    home.md
    services.md
    technical-audit.md
README.md

Example concept file

---
type: "Service"
title: "Technical Audit"
resource: "https://example.com/audit"
sources:
  - resource: "https://example.com/audit"
generated:
  by: "process:hota-okf-generator"
  at: "2026-10-17T00:00:00Z"
---
# Technical Audit

An overview of the service...

This is illustrative content. The generator uses actual public page text and source URLs. A generated timestamp documents extraction, not independent verification or the publication date of the source.

Why OKF can be useful for AI Search Optimisation

AI discovery and retrieval often work best when useful information has clear boundaries, contextual labels and recognizable connections. An OKF bundle makes those elements explicit. It can give compatible AI applications a cleaner source for knowledge retrieval, support version-controlled editorial review, and help teams maintain consistent definitions across services and products.

Discoverable concepts

Concept names and navigation links help an application locate the part of a knowledge collection relevant to a question.

Clearer comprehension

Individual Markdown documents reduce unrelated layout content and allow each topic to carry its own context.

Traceable sources

Source URLs show where extracted information originated. This is a foundation for editorial checks, not a guarantee of accuracy.

Portable knowledge

Markdown and YAML can be reviewed, edited, versioned and ingested by many existing tools without a custom data platform.

OKF does not automatically enroll a site in an AI search index. A bundle becomes useful to an external AI service only if that service can access it and chooses to consume it. Publication alone does not guarantee AI-generated citations.

What is new in OKF v0.2?

Version 0.2 improves the way a concept can communicate its origin, update history, verification and freshness. The sources field lists supporting materials. generated can indicate who or what created the document and when. Optional verified, status and stale_after fields add lifecycle context when appropriate.

These signals are especially relevant when knowledge is extracted or maintained by automated systems. They help distinguish an original source from a verified claim and make aging information easier to identify. The generator writes source and generation metadata, but it deliberately does not invent verification records or trust judgments.

How to generate and use your bundle

  1. Enter the homepage URL of a public website and choose up to 50 pages.
  2. The generator reads the entry page, identifies important internal destinations and requests their publicly available HTML.
  3. Each readable page becomes a Markdown concept with metadata, a source URL and extracted body text.
  4. Download the ZIP and open the index.md file to explore the concepts.
  5. Review descriptions, headings, relationships, factual claims, outdated content and permissions before publishing.
  6. Host or import the curated bundle into a knowledge system that explicitly supports reading Markdown files.

How to interpret a warning

Warnings such as missing headings or little extractable text indicate potential gaps in extraction quality. They are not proof that a page is inaccessible to every AI crawler. Some sites render significant content using JavaScript that this lightweight server-side generator does not execute.

How to improve the exported files

Replace vague descriptions with clear explanations, remove repeated navigation text, connect related concepts with Markdown links and record additional authoritative sources when you can verify them. The most valuable result is a maintained knowledge collection, not merely a ZIP download.

OKF compared with other AI-oriented files

FormatPrimary purpose
OKFPackages individual knowledge concepts, source information and relationships in portable Markdown files.
llms.txtProposes a concise entry point that directs models and people to useful site content.
Schema.orgExpresses structured information about entities and their properties, often in webpage JSON-LD.
sitemap.xmlLists URLs to assist website discovery and crawling.
robots.txtCommunicates crawler access rules for URL paths.

These formats address different needs. They can complement each other rather than serving as substitutes.

Original specification and technical references

Frequently asked questions

OKF is an open format for organizing concepts as Markdown files with YAML frontmatter. It makes knowledge portable, readable and easier for compatible software and AI systems to inspect.

This tool targets OKF v0.2 from the GoogleCloudPlatform knowledge-catalog specification. It produces a starter bundle for editorial review, not a formally verified knowledge corpus.

Enter a public website domain, choose a page limit up to 50, and generate. Review the extracted concept list and warnings before downloading the ZIP.

The generator begins with the entry page and prioritizes important internal links found on that page, such as service, product, about and documentation links. It does not claim to crawl an entire site.

The 50-page limit makes the free tool useful without placing excessive load on a website or the hosting server. Some sites return fewer eligible or accessible pages.

It includes an okf/index.md navigation file, separate concept Markdown files, YAML metadata, source URLs, generation timestamps and a README describing review requirements.

Frontmatter is a short metadata block at the start of a Markdown document. OKF v0.2 requires a nonempty type for each concept. Other fields such as title and description are recommended rather than universally required.

A concept is one Markdown file describing a piece of knowledge, such as a service, product, article, reference or business process. The file path acts as its concept identifier.

Start with the number of pages processed, then inspect the concept table. Missing headings or very limited text mean that a page may need editorial cleanup. Open index.md to navigate the generated files.

No. It extracts the publicly available webpage text and preserves source URLs. The generated material is not independently fact-checked or human-verified.

Provenance tells a reader or AI system where a concept came from. Generated files include the originating webpage as a source. Optional verification and trust fields are not fabricated.

Version 0.2 expands support for sources, generated and verified metadata, lifecycle and freshness fields, and attested computations. Those capabilities are optional and depend on real evidence.

No. OKF v0.2 makes index.md optional, but a navigable index is useful for progressively exploring larger knowledge bundles.

llms.txt is a proposed orientation file containing links and descriptions. OKF is a structured collection of individual Markdown knowledge concepts, metadata and cross-links.

No. Schema.org annotates entities and properties, often on web pages. OKF packages contextual knowledge into Markdown concepts. They can complement each other.

There is no guarantee. OKF may make knowledge easier to export, manage or retrieve when a system consumes it, but public AI search engines do not universally ingest OKF bundles as a ranking signal.

Review it first. Automatically extracted pages can include missing context, stale facts or irrelevant text. Verify sources, clarify concepts and confirm that you have the right to redistribute the content.

No. It requests public HTTP or HTTPS pages only, blocks private and reserved network addresses, and does not log in or execute browser JavaScript.

Generated ZIP files are made available through a random download token for up to 24 hours, subject to server storage policies.

Not necessarily. JavaScript-rendered websites, image-heavy pages and unusual templates may have less text visible to a server-side parser. Treat the warning as a prompt to review the extracted file.