Turn PDFs into source-anchored knowledge bundles
DocuShell exports parsed documents in the Open Knowledge Format — an open, vendor-neutral standard of Markdown with YAML frontmatter. Each bundle is structured, cross-linked, and tied to source provenance so AI agents and human reviewers can inspect what the bundle came from.
sha256:a1b2c3…What is OKF?
The Open Knowledge Format is an open specification for representing knowledge as a directory of Markdown files with YAML frontmatter. It was published by Google Cloud under the Apache 2.0 license and is intentionally minimal: if you can read a file, you can read OKF; if you can clone a repo, you can ship it.
DocuShell is a producer of conformant OKF v0.1 bundles. We turn the messy reality of PDFs into clean, typed, source-anchored knowledge any agent or human can consume.
- Human- and agent-readable — no SDK required
- Diffable and reviewable in version control
- Portable across tools, orgs, and time
- Open spec, Apache 2.0 — zero lock-in
From PDF to bundle in one request
OKF export rides on the same hardened parse pipeline that powers DocuShell's Markdown and JSON extraction.
Send a PDF
Call the Parse API with okf=true. Your file never leaves DocuShell's hardened pipeline.
Parse + anchor
DocuShell extracts structured Markdown, then Ethos checks parser evidence refs against the source fingerprint.
Get an OKF bundle
Download a conformant .okf.zip: concept docs, indexes, cross-links, and source-anchored provenance.
What's in a bundle
A self-contained directory of concepts. Reserved index.md files make it navigable; frontmatter makes it queryable.
acme-q3-earnings.okf.zip
├── index.md # bundle map + okf_version
├── log.md # generation history
├── documents/
│ ├── index.md
│ └── acme-q3-earnings.md # parsed content as an OKF concept
└── references/
├── index.md
└── source.md # source file + SHA-256 fingerprint---
type: Parsed Document
title: "Acme Q3 Earnings"
description: "Parsed from Acme Q3 Earnings.pdf by DocuShell parse-pdf."
tags: [docushell, parsed-document, okf]
source_fingerprint: sha256:a1b2c3…
pages: 8
evidence_anchoring: bound
okf_producer: docushell/parse-pdf
---
# Q3 Earnings
Revenue grew 18% YoY …
# Citations
[1] [Source document — Acme Q3 Earnings.pdf](/references/source.md)Source-anchored
Knowledge your agents can inspect, not just read
Anyone can convert a PDF to Markdown. DocuShell goes further when evidence anchoring is available: the Ethos layer checks parser evidence refs against the source document fingerprint and records the outcome right in each bundle. A consuming workflow can distinguish source-bound evidence from unverified derived text.
View Ethos on GitHubsource_fingerprintContent-addresses the exact file every concept came from.
evidence_anchoringPer-document verdict: bound, unverified, or skipped.
# CitationsLinks each concept back to its source of record.
Why teams ship OKF
Vendor-neutral by design
Plain Markdown + YAML frontmatter. No SDK, no proprietary API, no schema registry stands between you and your knowledge.
Drops straight into agents
Load bundles into any LLM context, RAG pipeline, or compatible knowledge catalog without a proprietary DocuShell reader.
Version-controllable
A bundle is a directory. Commit it to git for line-by-line diffs, pull-request review, and blame on your curated knowledge.
Source-anchored trust
The bundle carries the source fingerprint and DocuShell's Ethos anchoring verdict in frontmatter.
Progressive disclosure
Auto-generated index.md files let an agent navigate the hierarchy one level at a time instead of loading everything at once.
Works with your tools
Browse the same bundle in Obsidian, Notion, MkDocs, or any Markdown viewer — the format is just files.
Enable it with one field
Add okf=true to any Parse API request. The completed job exposes an okf_bundle_download link next to your Markdown and JSON artifacts.
curl -X POST https://api.docushell.com/v1/parse \
-H "Authorization: Bearer $DOCUSHELL_API_KEY" \
-F "[email protected]" \
-F "okf=true"
# When the job completes, the result includes:
# "okf_bundle_download": "/v1/jobs/<id>/download?format=okf_bundle"Frequently asked questions
What is the Open Knowledge Format (OKF)?
OKF is an open, vendor-neutral specification for representing knowledge as a directory of Markdown files with YAML frontmatter. It was published by Google Cloud under the Apache 2.0 license and is not tied to any model provider, database, or agent framework. DocuShell emits conformant OKF v0.1 bundles from your parsed PDFs.
How is an OKF bundle different from plain Markdown export?
A plain Markdown export is a single blob. An OKF bundle is a structured, self-describing knowledge package: each document is a concept with typed frontmatter, an index.md for progressive disclosure, cross-links between concepts, and source provenance tied to the source file by SHA-256 fingerprint.
What does “source-anchored” mean?
Each OKF bundle DocuShell generates is paired with a source concept that records the original file's SHA-256 fingerprint. When evidence anchoring is available, the bundle reports DocuShell's Ethos anchoring outcome so downstream systems can distinguish source-bound evidence from unverified derived text.
Will using OKF lock me into DocuShell or Google Cloud?
No. An OKF bundle is just a folder of Markdown files. You can read it with cat, browse it in Obsidian or Notion, render it with MkDocs, load it into any LLM context, or ingest it into a compatible knowledge catalog. The format is standalone and the spec is open, so there is no lock-in.
How do I get an OKF bundle?
Set okf=true on any Parse API request. DocuShell parses the PDF, anchors the evidence, and returns a downloadable .okf.zip bundle alongside the usual Markdown and JSON artifacts.
Ship source-anchored knowledge
Start free, then turn any PDF into an OKF bundle your agents can read with confidence.