Changelog

Title is now single-line with ellipsis truncation on overflow.

Layout order changed from title/date/tags to title/tags/date

for better scannability.

Date now shows "February 28, 2026 · 12:29 PM" instead of just

"February 28, 2026" in the article detail pane.

NormalizeContentFromJson intercepts ProseMirror-style JSON documents

({"type": "doc", "content": [...]}) that the AI hallucinates instead

of markdown text, converting them back to markdown before block

parsing. Handles headings, paragraphs, bullet/ordered lists, code

blocks, blockquotes, tables, and horizontal rules via recursive node

tree extraction.

Intent slot descriptors for notes.create_note and notes.update_note

now explicitly warn against JSON document objects in the content slot.

10 new NUnit tests cover plain markdown passthrough, ProseMirror doc

conversion, tables, bullets, code blocks, mixed types, invalid JSON,

non-document JSON, empty strings, and deeply nested table cells.

ParseActionBlocks now falls back to ExtractBareIntentJson when no

[ACTION] wrapper tags are found, recovering intent JSON that the AI

emits without the required tags. The fallback only triggers when the

primary parser found nothing, preventing double-extraction.

ActionFormatHeader now explicitly warns against JSON document objects

in content/description text slots.

Visibility changes: ParseActionBlocks, ExtractBareIntentJson,

ParsedAction, and FlattenSlotValue are now internal (testable via

InternalsVisibleTo).

6 new xUnit tests cover bare JSON recovery, surrounding prose

stripping, non-intent JSON rejection, multiple intents, and

malformed JSON handling.

Refactor FetchFeedAsync to follow rel="next" pagination links,

retrieving up to 20 pages of older entries during initial feed

import. This allows the RSS plugin to fetch the full history from

paginated feeds (like the PrivStack changelog) instead of only the

first page. RefreshFeedAsync is unchanged — it only needs the

latest page for new article detection.