Move status bar to outer grid to fix content overlap
Details
The status bar was nested inside the content column grid alongside
plugin views. Despite using RowDefinitions="*,Auto", the plugin
content was not being properly constrained above the status bar,
causing the bottom of scrollable plugin views (Finance reports, budget,
etc.) to be clipped behind the status bar.
Restructured the layout so the status bar is in Row 3 of the outer
window grid (RowDefinitions="Auto,Auto,*,Auto"), completely outside
the sidebar+content area. The content column is now a simple Panel
with ClipToBounds, and the status bar spans the full window width
below both the sidebar and content area. This guarantees the content
area is always constrained above the status bar.
Add bottom margin to plugin content area to prevent status bar overlap
Details
Plugin views were rendering flush against the bottom status bar with no
spacing, causing the last items in scrollable content to appear clipped
or overlapped by the status bar. Added a 4px bottom margin to the
ContentControl that hosts plugin views in MainWindow.axaml, so all
plugins automatically get breathing room above the status bar.
Add bottom padding to Dashboard Overview ScrollViewer
Details
The Overview tab's ScrollViewer had insufficient bottom padding (24px),
causing the last plugin card row to be clipped behind the status bar.
Increased to 32px to match the spacing used elsewhere. The Data tab
already had adequate padding (80px).
Increase local model token and sentence limits
Details
Medium tier responses were getting cut off mid-sentence (e.g.
actionable budget advice truncated at item 3). The system prompt
already instructs the model to be concise, so the hard limits
were redundant safety nets causing premature truncation.
Bumped local model limits:
- Short: 60→80 tokens, 2→3 sentences
- Medium: 250→400 tokens, 5→8 sentences
- Default: 150→200 tokens, 3→4 sentences
Mistral 7B has sufficient context window for these increases.
Bumped desktop to 1.60.3.
Improve Duncan AI context for Finance and local models
Details
Three changes to fix Duncan giving poor/irrelevant responses when
users ask about their budget:
1. Increased RAG chunk limits for local models (Mistral): result
limit 6→8, max chunk chars 400→800. Mistral's 8K context can
handle this and the budget data was being severely truncated.
2. When FetchActiveItemEntityAsync fails to find an entity via SDK
or DataSourceProvider (e.g. finance_budget synthetic types), fall
back to formatting InfoPanelDetailField entries into the active
item context instead of leaving it as just a title string.
3. Bumped desktop to 1.60.2.
Get notified about new releases