This document guides Seed Agents on behavior

Seed Hypermedia is a decentralized network of connected Documents and Comments, both are resources identified with hm://<ACCOUNT_ID>/<PATH> urls.

All Seed data is saved into IPFS blobs, with data encoded as IPLD DAG-CBOR

Documents

Seed Documents are organized in a hierarchy with path names. Logically this behaves like a file system where each folder is a document (or has an index).

Each document is also a hierarchy of blocks, within the content field. Documents are managed by creating operations which are packed into Change blobs. The changes reference previous changes, and all the operations are walked to reach the current document state. Refs are used for naming the document and locating it, and they can reference any change.

Blocks include Paragraph and Heading which constitute the text material.

Docs can embed other Resources

Documents also support arbitrary metadata with the SetAttributes operation.

Comments

Comments have a "TSID", an auto-generated ID that acts as a non-human-readable path.

They are edited by creating a new comment blob, and deleted by editing the comment to be empty.

Importing

When importing you must follow the hierarchy convention of documents with human-readable path names, and each document must be correctly organized

Imported content must include the importedAsset metadata field which references the source material that is archived to ipfs with UnixFS, so the field will contain the ipfs:// URL so that readers can view the source content.

Import Web

If importing a website, archive the content as raw HTML+Assets, and save to the asset field.

Importing Markdown

Follow document conventions so documents have the correct hierarchy.

Importing PDF

Follow the TODO:Guide to convert the PDF content to markdown within your memory. Then follow the Markdown import instructions.