Sources
Each topic has a local source library. Open the workspace inspector, choose Source type, give the item a title, and select Add source.
Supported source types
Section titled “Supported source types”- Pasted text: stored as a readable
.txtfile. - Local text file:
.md,.markdown, and.txtfiles up to 2 MiB. Markdown stays Markdown; line endings are normalized for portability. - URL reference: stores the complete
http://orhttps://address in a small Markdown reference file. Dusori does not fetch or copy the page. - Accepted research: stores a Microsoft Learn catalog reference or an English Wikipedia plain-text extract only after preview and explicit acceptance.
URLs containing embedded usernames or passwords are rejected. Opening a saved URL is an explicit browser action and can contact that website.
Research from a roadmap objective
Section titled “Research from a roadmap objective”The Research section defaults to the next unchecked objective. Search Microsoft Learn ranks the public module catalog locally. Search Wikipedia uses the English search API; previewing a result retrieves that selected page’s plain-text extract. Each provider is blocked on first use until you accept its host-specific disclosure. Nothing else from the workspace is sent.
Results remain suggestions until you choose Add to sources. Microsoft Learn captures are labeled as catalog references, not page snapshots. Wikipedia extracts stay below the same 2 MiB source cap and end with [truncated] when the full extract would exceed it. Dismiss records the result key locally so it stays out of later searches.
Topic file contract
Section titled “Topic file contract”Topics/<topic-slug>/ Sources/ manifest.json items/ <sha-prefix>-<portable-title>.md|txt research.json # created after the first dismissal Updates/YYYY/MM/YYYY-MM-DD.mdmanifest.json records the capture method, SHA-256 hash, local path, media type, byte size, timestamp, and optional original filename, URL, or research origin. Research origin names mslearn or wikipedia, the capture method, and the capture time. URL sources—including research captures—deduplicate by canonical URL. A successful new capture also appends a line to the topic’s dated update log.
Source files and research.json are included in workspace ZIP exports and remain ordinary readable files when the Dusori root sits inside an Obsidian vault. Arbitrary web-page fetching, PDF extraction, key-based search, and AI transformation remain planned work.
Applying a curriculum import also stores the pasted official outline here before updating the topic roadmap. Re-importing identical outline text reuses the existing source record.