Changelog

Recording

• A new memory backend for recording improves how audio data is retained during sessions
• In-memory recordings are now preserved when audio degradation occurs, so you won't lose what was already captured

Chat

• The AI button now has a label for better discoverability
• Chat UI components have a cleaner white background and improved layout stability
• The chat window no longer expands off-screen when opening the panel

UI

• A new meeting button appears at the end of the header for quicker access
• Text in event descriptions is now selectable
• The listen button blends with the background color
• Fixed tab visual state and editor tab behavior

Settings

• Fixed an issue where excluded apps for mic detection in notification settings were not applied correctly

OTA

• Add app name migrater for macOS

Transcription

• Various live transcription bug when using our PRO proxy are fixed
• Batch transcription is more reliable for local models and multi-channel recordings
• Failed transcript runs now show clearer errors when you retry

Search

• Search now lives in the left sidebar, so you can move between recent activity and search results without leaving the main view

Storage

• You can now choose a custom location for Char content in Settings
• Obsidian vaults are listed as a storage destination, so you can point Char directly at your vault
• When moving your content, Char can copy existing sessions to the new folder before restarting

Chat

• The floating chat window can now resize freely without clipping or layout issues

Folders

• Moving and renaming folders is now more reliable, with file system changes applied before updating app state

Timeline

• The Today timeline places the current-time marker more accurately, including during ongoing events and in your selected timezone

Search

• Search now lives in the left sidebar, so you can move between recent activity and search results without leaving the main view

Storage

• You can now choose a custom location for Char content in Settings
• When moving your content, Char can copy existing sessions to the new folder before restarting

Transcription

• Batch transcription is more reliable for local models and multi-channel recordings
• Failed transcript runs now show clearer errors when you retry

Timeline

• The Today timeline places the current-time marker more accurately, including during ongoing events and in your selected timezone

Calendar

• Google Calendar events can now sync into Char alongside Apple Calendar (Beta)
• Calendar sync is more reliable when calendars are added, removed, or toggled in settings

Notes

• Search and replace now works more reliably across notes and transcripts
• Session preview cards now show richer previews, including enhanced notes and Markdown formatting when available

Chat

• Chat now suggests one-click starter prompts like summaries, action items, follow-up emails, key decisions, and main topics

Transcription

• Batch transcription is more reliable, with fixes for provider errors and edge cases that could leave a session stuck
• Re-running a transcript no longer clears the old transcript before the new run actually begins
• Added Parakeet support in Cactus as a new local transcription model option

Folders

• Folders tab now shows a cleaner placeholder state while we prepare a larger update for organizing notes

Transcription

• Batch transcription is more reliable, with fixes for error handling and edge cases that could leave a session stuck

Account

• Sign-out and account deletion are now grouped together in one section of account settings

Folders

• Folders tab shows a clean placeholder state while we work on a new design for organizing your notes

Calendar

• Google Calendar events can now sync into Char alongside Apple Calendar
• Calendar sync is more reliable when calendars are added, removed, or toggled in settings

AI

• Added Parakeet support in Cactus as a new local transcription model option

Notes

• Search and replace now works more consistently across notes and transcripts
• Session preview cards now show richer previews, including enhanced notes and Markdown formatting when available
• You can keep working in other note formats while batch transcription runs, with progress shown inline in the Transcript tab
• Re-running a transcript is now safer and avoids clearing the old transcript until the new run starts

Chat

• Chat now suggests one-click starter prompts like summaries, action items, follow-up emails, and key decisions

Calendar

• URLs in calendar event descriptions are now rendered as clickable links

Chat

• Chat now shows which session it has as context via a context indicator
• Chat can now directly edit your notes — ask it to revise, reformat, or expand content in place

Notes

• Code blocks are now properly supported in the editor
• Mentions are now exported as plain text in Markdown

Export

• Added Org mode (.org) as an export format alongside PDF, TXT, and Markdown

AI

• Short tips are now shown while the AI is generating a summary

Other fixes and improvements

• Failed batch transcription no longer resets the in-progress transcription

Chat

• Various improvements made for the next stable release

Export

• Added Org mode (.org) as an export format alongside PDF, TXT, and Markdown

Calendar

• URLs in calendar event descriptions are now rendered as clickable links

Mostly internal changes and small bug fixes.

AI

• Pre-meeting notes are now passed to the AI as a separate section — summaries better reflect what actually changed during the meeting versus what was prepared beforehand
• Azure OpenAI and Azure AI Foundry are now supported as LLM providers (beta)
• Enhance (summarize) now stays active correctly when switching between tabs

Chat

• Chat is now a persistent panel instead of a floating overlay — stays in context as you work

Notes

• Hovering over a session in the sidebar or tab bar now shows a preview card with its title, date, participants, and a text snippet
• Typing @ with no query now shows default suggestions — sessions, people, and organizations — instead of an empty dropdown
• Long mention names are truncated so they don't overflow the editor
• Links now open on hover instead of requiring Cmd+Click

Export

• Export options are consolidated into a single Export button that opens a modal — choose format (PDF, TXT, Markdown) and what to include (summary, transcript)

Audio

• Playback speed is now configurable directly in the transcript view

Integrations

• Connecting integrations can now be triggered directly from the desktop app

Other fixes and improvements

• Detect notifications setting is now properly respected
• List items and task items are guaranteed to start with a paragraph — fixes edge cases in structured note editing

AI

• Pre-meeting notes are now passed to the AI as a separate section
  • So summaries better reflect what actually changed during the meeting versus what was prepared beforehand
• Azure OpenAI and Azure AI Foundry are now supported as LLM providers (beta)
• Enhance(summarize) now stays active correctly when switching between tabs

Notes

• Hovering over a session in the sidebar or tab bar now shows a preview card with its title, date, participants, and a text snippet
• Typing @ with no query now shows default suggestions — sessions, people, and organizations — instead of an empty dropdown
• Long mention names are truncated so they don't overflow the editor
• Links now open on hover instead of requiring Cmd+Click

Export

• Export options are consolidated into a single Export button that opens a modal
  • choose format (PDF, TXT, Markdown) and what to include (summary, memos, transcript)

Audio

• Playback speed is now configurable directly in the transcript view

Settings

• Detect notifications in the settings is now properly respected

Transcription

• Fixed transcript segmentation producing incorrect splits between speakers
• Smarter mic detection: when mic usage is detected, the notification now lists nearby calendar events so you can jump straight into the right session

Notes

• Link text edits now automatically update the underlying URL — no more stale hrefs after renaming a link
• Links no longer bleed into adjacent typed text when editing near a link boundary
• Pressing Enter on an empty line now clears active inline formatting (bold, italic, underline)
• Link hover tooltip is now correctly positioned for links near the top of the editor

Audio

• Recordings are now saved as MP3 files

Calendar

• Session creation from a calendar event now correctly populates participants from event attendees
• Calendar onboarding now shows a Request Access to Calendar button instead of a dead-end "No calendars found" message

Onboarding

• Redesigned onboarding flow with clearer step states, checkmark completion indicators, and auto-scroll navigation between steps

Auth

• Fixed sign-in flow when the user was already logged into char.com in their browser

Other fixes and improvements

• Fixed red artifact line appearing in the left sidebar timeline
• Fixed account page side borders not extending to full height
• "Resume listening" now appears in the overflow menu when a transcript already exists, instead of "Start listening"
• AI generation is now more resilient when output validation retries exhaust — gracefully yields instead of erroring out
• Refreshed button styles with improved visual depth and hover feedback

Transcription

• Fixed transcript segmentation producing incorrect splits between speakers
• Smarter mic detection: app now recognizes when a calendar event is recently active when prompting to record

Notes

• Link text edits now automatically update the underlying URL — no more stale hrefs after renaming a link
• Pressing Enter on an empty line now clears active formatting (bold, italic, underline)

Audio

• Recordings are now saved as MP3 files

Calendar

• Session creation from a calendar event now correctly populates participants from event attendees
• Calendar onboarding now shows a Request Access to Calendar button instead of a dead-end "No calendars found" message

Onboarding

• Redesigned onboarding flow with clearer step states, checkmark completion indicators, and auto-scroll navigation between steps

Auth

• Fixed sign-in flow when the user was already logged into char.com in their browser

Other fixes and improvements

• Fixed red artifact line appearing in the left sidebar timeline
• Fixed account page side borders not extending to full height
• AI generation is now more resilient when output validation retries exhaust — gracefully yields instead of erroring out

Transcription

• Whisper small via Cactus engine is now in beta — run speech-to-text locally on your Mac (English only)
• Smarter language-aware provider selection: the best STT backend is now chosen automatically per language
• Fixed visual glitch when starting a new listening session
• "Resume listening" button label now correctly reflects the paused state
• Fix mic detection being skipped when the app window is focused
• Fix blank speaker label in transcripts when user name is not set

AI

• Your selected default template in settings is now applied automatically to all new AI-generated summaries
• AI-generated summaries now produce section headings in the correct language for non-English users
• AI model selectors filter out deprecated and non-chat models — cleaner model list
• Fix Gemini 2.0 and 2.5 models (Flash, Pro) incorrectly being hidden from the AI model list
• Local provider status (Ollama, LM Studio) shown with visual indicators — model selection disabled when provider is not running
• New Memory tab in AI settings for custom vocabulary

Search

• Transcript search now supports replace, match case, and whole word options with keyboard shortcuts
• Search auto-scrolls to the active result and stays in sync between transcript and editor views

Calendar

• Improved calendar visual hierarchy — cleaner header spacing, typography, and weekend styling
• Calendar onboarding now shows a loading state while fetching calendars
• Improved participant email resolution for Apple Calendar events
• Long event descriptions no longer overflow in metadata chips and popovers

Notes

• Cmd+Click on links in the note editor now opens them in your default browser
• Added "Reveal in Finder" to the session overflow menu

Contacts

• Contact avatars now show colored initials when no profile photo is set

Onboarding

• Streamlined onboarding flow with clear trial status after sign-in
• Removed contacts permission request from onboarding

Other fixes and improvements

• Renamed remaining "Hyprnote" references to "Char" across the UI
• Settings and AI tab bars now wrap correctly on narrow windows
• Fixed hover jump on interactive buttons
• Telemetry opt-out toggle now properly disables usage analytics
• Fix STT provider validation for cloud meta-models — routing now resolved server-side
• Cmd+Q quit now triggers on keydown for more predictable behavior

AI

• Your selected default template in settings is now applied automatically to all new AI-generated summaries
• Whisper small support by Cactus is now beta. English only.

Transcription

• Fixed a visual glitch that occurred when starting a new listening session
• "Resume listening" button label in the header now correctly reflects the paused state

Notes

• Added "Reveal in Finder" to the session overflow menu for quick file access

Onboarding

• Help users to focus on current step
• Show clear trial status after sign-in

Others

• Settings and AI tab bars now wrap correctly on narrow windows
• Long calendar event descriptions no longer overflow their container

Transcription

• Fix mic detection being skipped when the app window is focused
• Fix blank speaker label in transcripts when user name is not set

AI

• AI-generated summaries now produce section headings in the correct language for non-English users

Contacts

• Contact avatars now show colored initials when no profile photo is set

Notes

• Cmd+Click on links in the note editor now opens them in your default browser

Others

• Renamed remaining "Hyprnote" references to "Char" across the UI
• Ensure test data fixtures do not leak into release builds

AI

• Fix Gemini 2.0 and 2.5 models (Flash, Pro) incorrectly being hidden from the AI model list
• Experimental Cactus inference engine for speech-to-text

Others

• Linux ARM64 (aarch64) builds are now available (Although Linux support is still experimental)

UI

• Transcript search now supports replace, match case, and whole word options, with keyboard shortcuts and improved navigation between matches
• Search now auto-scrolls to the active result and stays in sync between transcript and editor views
• Calendar UI has improved visual hierarchy (cleaner header spacing/typography and clearer weekend styling)
• Calendar onboarding now shows a proper loading state while calendars are being fetched, and onboarding button overflow spacing has been fixed
• Fixed metadata panel scrolling behavior in session headers for more consistent layout handling

Transcription

• Fix bugs where meta-models (e.g. cloud) were incorrectly validated against a specific provider — provider routing is now resolved server-side based on language support
• Improved language-aware provider selection: the best STT backend (Deepgram vs. Soniox) is now chosen automatically per language

AI

• AI model selectors now filter out deprecated models (GPT-3.5, Claude 2, etc.), date-snapshot versions, and non-chat models (codex, moderation, realtime)
• Local provider connection status (Ollama, LM Studio) is now shown with visual indicators — model selection is disabled when the provider is not running

Calendar

• Improve participant email resolution for Apple Calendar events with a better fallback chain (email → contact → URL → name)

Other fixes and improvements

• Fix Cmd+Q quit behavior — now triggers on keydown instead of keyup for more predictable behavior
• Status banner position adjusts dynamically based on timeline visibility
• Profile avatars now use Facehash-generated faces
• Fix hover jump on interactive buttons caused by gradient backgrounds
• Remove contacts permission request from onboarding

Notification

• Old MicStarted event and experimental MicProlongedUsage event are merged
• Now accept threshold configuration, which decides how long the mic must be active before triggering

AI

• Add Mistral as a new transcription provider
• Auto-trigger summary generation after transcript upload or batch completion

Search

• Full-text search now powered by Tantivy, faster indexing and more accurate results

Chat (experimental)

• Bring back chat as an experimental feature with major improvements
• Richer session context passed to chat for more relevant responses
• Search tool results now include full session context

Transcription

• Live transcription failure is now "non-fatal", recorder keeps running, entering degraded mode instead of crashing the session
• Show degraded transcription status in the transcript view

Calendar

• Improve calendar event sync reliability across timeline and notifications
• Show event descriptions in session metadata chips

Notes & Timeline

• Multi-select sessions in timeline with bulk delete and undo toast
• Auto-focus title input when no title is set
• Remove live session tab pinning

Other fixes and improvements

• Fix desktop auth refresh edge case tied to window visibility events
• Replace onboarding window with new onboarding within the main window
• Preserve preferred audio input device priority across device reconnects
• Improve audio import experience and error handling for AAC-in-MP4 files
• Fix fullscreen-to-hidden transition causing black screen on macOS
• Fix PDF export escaping for special characters in markdown
• Fix pro model configuration when starting trial from settings
• Restore relaunch after system audio permission grant

Notification

• Old MicStarted event and experimental MicProlongedUsage event are merged
• Now accept threshold configuration, which decides how long the mic must be active before triggering

AI

• Add Mistral as a new transcription provider
• Auto-trigger summary generation after transcript upload or batch completion

Search

• Full-text search now powered by Tantivy, faster indexing and more accurate results

Chat (experimental)

• Bring back chat as an experimental feature with major improvements
• Richer session context passed to chat for more relevant responses
• Search tool results now include full session context

Transcription

• Live transcription failure is now "non-fatal", recorder keeps running, entering degraded mode instead of crashing the session
• Show degraded transcription status in the transcript view

Calendar

• Improve calendar event sync reliability across timeline and notifications
• Show event descriptions in session metadata chips

Notes & Timeline

• Multi-select sessions in timeline with bulk delete and undo toast
• Auto-focus title input when no title is set
• Remove live session tab pinning

Other fixes and improvements

• Fix desktop auth refresh edge case tied to window visibility events
• Replace onboarding window with new onboarding within the main window
• Preserve preferred audio input device priority across device reconnects
• Improve audio import experience and error handling for AAC-in-MP4 files
• Fix fullscreen-to-hidden transition causing black screen on macOS
• Fix PDF export escaping for special characters in markdown
• Fix pro model configuration when starting trial from settings

Chat

• Replace feedback modal with support chat — report bugs and suggest features via chat instead of the feedback form

Search

• Collapse search bar to icon by default, expand on click
• Advanced search tab with filters (Meeting note, Person, Organization) and date presets (Today, This week, This month)

Keyboard Shortcuts

• Restore Cmd+Shift+Q for Quit Completely, fix app menu quit handler

UI & Experience

• Context menu for creating new notes and tabs (right-click in main area)
• macOS dock right-click menu now shows "New Note" for quick access
• Fix tab bar blur visibility and set tab width to 160px
• Tray settings menu labels match in-app: "Open AI" → "AI Settings", "Open General" → "App Settings"

Settings

• Polish settings UI — simplify labels, timezone selector, week start, and lab settings
• Download button with dynamic channel selection for updates
• Add link to local LLM setup guide in AI settings

AI & Transcription

• Fix model reset when switching STT providers
• Fix delete model command for local STT
• Handle BCP47 language codes in summary template filters

Fixes

• Fix export transcript timestamp offset for multi-transcript sessions
• Fix auth session handling — separate Supabase sessions for desktop to prevent refresh token competition
• Fix store initialization racing with persisters — prevents "Welcome to Char" session from disappearing on startup
• Fix sign-in waiting screen — always show callback URL input inline
• Fix empty sessions and template tab state transitions
• Fix calendar timeline events (ignored, all-day types)

Calendar

• Redesigned calendar view with week navigation, responsive grid, and event popovers
• All-day events display and week start configuration

Chat

• Replace feedback modal with support chat — report bugs and suggest features via chat
• Chat can now open as a full tab
• Slash commands in chat input for quick actions

Search

• Collapse search bar to icon by default, expand on click
• Advanced search tab with filters (Meeting note, Person, Organization) and date presets (Today, This week, This month)

Keyboard Shortcuts

• Cmd+Shift+Q to Quit Completely (restored)
• Cmd+Shift+N to create a new note and start listening
• Cmd+O now works when search bar is focused on empty tab

UI & Experience

• Context menu for creating new notes and tabs (right-click in main area)
• macOS dock right-click menu now shows "New Note"
• Templates and chat shortcuts accessible as tabs
• Tray menu labels now match in-app: "AI Settings", "App Settings"

Settings

• Polish settings UI — simplify labels, timezone selector, and lab settings
• Download button with dynamic channel selection for updates
• Add link to local LLM setup guide in AI settings

Fixes

• Fix STT model selection resetting unexpectedly; fix local model deletion
• Fix export transcript timestamp offset for multi-transcript sessions
• Fix auth session handling — separate Supabase sessions to prevent refresh token competition
• Fix store initialization racing with persisters — prevents "Welcome to Char" from disappearing on startup
• Fix sign-in waiting screen — always show callback URL input inline
• Fix empty sessions and template tab state transitions
• Fix calendar timeline events (ignored, all-day types)
• Fix BCP47 language codes in summary template filters
• Improve Zoom mute detection using native macOS Accessibility API

Keyboard Shortcuts

• New Cmd+Q behavior - hold to quit overlay (inspired by Chrome), replacing the previous hide behavior

Notifications

• Fix notification macOS glass border appearance
• Experimental in-meeting-reminder notification - opt-in in lab settings, notifies when another app has been using your microphone for 3+ minutes while Char isn't listening

Calendar

• New calendar view redesign - full calendar interface with week view, event popovers, and improved navigation controls
• Context menu to unignore calendar events directly from the sidebar timeline
• Fix refresh after-meeting rotation

Note Management

• Cmd+Shift+N hotkey to create a new note and start listening in one action
• Fix opening a note now switches to existing tab instead of creating a duplicate
• Fix markdown conversion regression causing note content issues

Mentions & Search

• Implement backlink mentions for notes, people, and companies with type badges and improved styling
• Fix: enable Cmd+O hotkey when search bar is focused on empty tab

UI & Experience

• Fix custom checkbox styling for MDX task lists in prose

• New cmd+q behavior inspired by Chrome. cmd+shift+q is removed
• Fix notification macOS glass border appearance
• Add version number to Open Char tray menu item
• Use dynamic product name for Main window title in Mission Control
• Implement experimental backlink mentions for notes, people, and companies

• Add experimental in-meeting-reminder notification.
  • Opt-in in the lab settings. Not tested much yet.
  • This will notify you when another app has been using your microphone for 3+ minutes while Char isn't listening.
• Add context menu to unignore calendar events in the sidebar timeline
• Add fallback for local AI models without tool calling support (e.g., Gemma via LM Studio)
• Fix opening a note now switches to existing tab instead of creating a duplicate
• Fix markdown conversion regression causing note content issues
• Fix crash when removing corrupted model file after checksum mismatch
• Fix empty sessions left behind when closing tabs
• Fix potential crash on startup related to log directory initialization
• Various onboarding fixes and simplification
• Fix bug report submission failing for long reports

Note Management

• Undo delete with 5-second countdown timer - accidentally deleted notes can be restored via the "Restore" button or Cmd+Z, with a dissolving animation showing the remaining time
• Space hotkey for controlling audio player playback
• Recently opened notes now appear in a "Recent" section at the top of the Open Note dialog (Cmd+O) for quick access

Settings

• Manual timezone configuration - override the system timezone specifically for how meeting times appear in the sidebar timeline, with 18 common timezone presets
• AI settings shortcut changed to Cmd+Shift+Comma
• New searchable dropdown component for language and timezone selection

Calendar

• Fix duplication of rescheduled recurring calendar events - previously, rescheduling a recurring event instance would show both old and new occurrences

AI & Transcription

• Auto-enhance now correctly counts words across multiple transcript segments, fixing cases where short segments wouldn't trigger enhancement
• LM Studio context length errors now show a "Learn how to fix this" link pointing to documentation with step-by-step instructions

UI & Experience

• Tooltip with keyboard shortcut hint for sidebar toggle button
• Hide retranscribe button during active listening to reduce visual clutter
• Improved dropdown keyboard navigation with Tab key support and visual selection indicator
• Dropdown selections auto-scroll to keep the selected item visible

Stability & Performance

• Fix memory leaks from actor subscriptions and timer accumulation not being cleaned up
• Fix high CPU usage caused by MutationObserver watching the entire document tree - now uses targeted ResizeObserver
• Fix transcript data migration to preserve speaker hints and support multiple transcripts per session

System

• Tauri 2.10.1
• Major dependency updates: cpal 0.17 (audio, now using upstream instead of patched fork), reqwest 0.13, cidre 0.14 (macOS bindings), tokio-tungstenite 0.28

• Add undo delete with visual feedback: deleted notes show dissolving animation with 5-second countdown timer, hover to pause and reveal "Restore" button, or press Cmd+Z to instantly restore
• Add manual timezone override in Settings > Language & Region for sidebar timeline display
• Add recently opened sessions tracking for quick access to recent notes
• Add space hotkey for controlling audio player playback
• Add tooltip with keyboard shortcut for sidebar toggle button
• Rename "Language & Vocabulary" to "Language & Region" in settings
• Improve subscription status display: show detailed status (PRO, trialing with days remaining, past due, canceled) with urgent styling when trial has 3 or fewer days
• Improve onboarding flow: new final step that automatically starts Pro trial for eligible users
• Improve context menu text consistency (capitalized "Open in New Tab", "Delete Note", etc.)
• Fix "invalid Refresh Token: Already Used" error in authentication
• Fix auto-enhance logic with proper eligibility checks (minimum 5 words required)
• Fix transcript data migration to preserve speaker hints and support multiple transcripts per session
• Fix duplication of rescheduled calendar events
• Fix memory leaks and high CPU usage
• Hide retranscribe button during active listening
• Improve scroll behavior for dropdown selections
• Remove trial begin/expired modals (replaced with inline status display)
• Various upstream updates (Tauri 2.10.1)

• Add "Open Note" action with Cmd+O shortcut to quickly search and open existing notes
• Various upstream updates containing critical fixes (unicode handling during markdown conversion, device enumeration segfault, etc.)

Navigation

• Open Note command (Cmd+O) with search palette to quickly find and open existing notes
• Pinned tabs now persist across app restarts

Transcript

• Improved empty transcript state with clearer visual feedback

Feedback

• Select issue type (bug report, feature request, question) when submitting feedback

UI & Experience

• Meeting icons (Zoom, Meet, Teams, Webex) now load locally for better reliability
• Tray menu "Hide" renamed to "Close All" for clarity

System

• Fixed stale single-instance socket that was preventing app launch on macOS
• Bumped markdown-rs to 1.0, which fixes a panic during markdown conversion containing a Unicode character

• Smarter mic notification filtering to reduce noise
• Fix app startup issue on macOS when stale socket exists
• Pinned tabs now persist across app restarts

• Add confirmation popover when closing active tabs
• Sidebar automatically expands when search is triggered
• Improve feedback submission flow with direct log attachment
• Video chat link visibility adapts to editor width
• Fix timeline sorting with fallback to event start time
• Only show changelog after app update, not on first install
• Rename tray menu 'Hide' to 'Close All' for clarity

Audio & Recording

• Acoustic Echo Cancellation (AEC) eliminates the need for heuristics-based micOnly mode
• Record microphone and speaker as separate stereo channels for cleaner audio separation
• Recordings saved as .wav instead of .ogg for better compatibility
• Sleep detection automatically stops recording when your laptop lid closes or system sleeps
• Audio visualizer now uses RMS-based amplitude for smoother, more accurate loudness display
• Improved macOS speaker capture with real-time safe buffer reuse
• Device priority selector in audio settings
• Fixed null pointer crash in microphone detection on macOS 26.x

AI & Speech-to-Text

• Added Mistral as a new LLM provider
• Added ElevenLabs as a new speech-to-text provider
• Language support checking warns when your selected STT provider doesn't support your language
• Show local model in Finder and delete downloaded models from AI settings
• Improved automatic model selection for Char Cloud users
• Better Deepgram language variant recommendations
• Fixed custom OpenAI endpoint configuration not being respected

Editor

• Added table support for inserting and editing tables directly in notes
• Added search and replace with case sensitivity option
• Cmd/Ctrl+click to open links in new tab
• Search auto-populates from selected text
• Improved markdown rendering styles and spacing

Attachments

• Images saved as files in session folder instead of base64
• Duplicate filenames automatically get numbered suffixes

PDF Export

• Export beautiful PDFs with metadata cover page including title, date, and participants
• PDFs save directly to Downloads folder and open automatically

Transcript

• Regenerate transcript from existing audio files via note header
• Optimize transcript word rendering and scrolling performance
• Fixed partial words leaking to inactive session transcripts during live recording

UI & Experience

• Redesigned app settings with improved layout and UX
• Changed app font from Inter to SF Pro
• Recording indicator overlay on dock icon while recording
• Split floating action button into separate "Join" and "Start listening" buttons
• Listening tab pinned to fixed left position during recording
• Sidebar automatically expands when searching
• Sidebar auto-scrolls to "Now" indicator on initial load
• Expandable fields in metadata panel
• "Open in new tab" context menu option
• Confirmation dialog when closing active listening tabs
• Video chat link visibility adapts to available editor width
• Event start times displayed in timeline sidebar
• Fixed timeline bucket sorting (weeks before months, correct chronological order)
• Reduced toast notification delays

Contacts

• Pinned contacts with drag-and-drop reordering
• Duplicate contact detection with merge functionality
• Current user pinned at top with "Me" badge
• Preserve "me" label when merging contacts

Calendar

• Enhanced context menu: "Open in new tab", "Reveal in Finder", "Delete completely"
• Create calendar events from Apple Calendar integration
• Fixed recurring event sync and recurrence series ID population
• Fixed timezone offset causing dates to appear off by one day
• Ignored events now persist after app restart

Data & Storage

• Replaced sqlite-based persisters with json and markdown-based persisters

Authentication & Account

• Auth session persists across app updates (no more re-login after updates)
• Stay logged in when app is backgrounded
• Fixed authentication token refresh errors
• Pro trial auto-starts on first app launch
• Prevent trial expired modal from interrupting onboarding
• Show "Upgrade to Pro" for expired trials instead of disabled features
• Auto-reset cloud STT and Char LLM settings when trial expires

Feedback & Support

• In-app feedback modal for bug reports and feature requests via Help menu
• Report Issue button on error page with pre-filled device info
• Git hash display in feedback modal

System

• Cmd+Q intercepted by default; Cmd+Shift+Q to force quit
• Tray menu shows update download progress
• Handle menubar "New Note" navigation correctly
• Reordered onboarding permissions for better UX flow

• Fix null pointer crash in mic detection
• Rust based legacy database loader for faster startup
• Prevent trial expired modal from interrupting onboarding flow
• Redesign app settings with improved layout and UX
• Add option to proceed without account during onboarding

This is an empty release caused by an automation mistake.

• Add echo cancellation. No more heuristics-based micOnly mode which causes confusion
• Update PDF export to include metadata cover page
• Make cmd+q be intercepted by default, and introduce cmd+shift+q to force quit the app
• Save recordings as .wav instead of .ogg for better compatibility with webkit
• Show microphone and speaker as overlays in audio player inside transcript view
• Reorder onboarding permissions for better UX flow
• Reflect update download progress state in tray menu
• Handle /app/new path in navigation events for menubar New Note
• Add feature flags plugin, and hide chat button behind it
• Migrate to streamdown v2 for AI responses
• Fix custom OpenAI endpoint configuration not being respected
• Implement rollback for partial folder move operations
• Improve filename generation and conflict handling
• Add loading spinner when refreshing plan status in account settings
• Show "content" and "others" paths in storage settings

• Add table support in editor

  • Insert and edit tables directly in your notes
  • Tables render with clean styling

• Fix authentication when app is backgrounded

  • Token refresh now stays active even when the app loses focus

• Support MP4 files in audio import

  • MP4 files can now be imported alongside other audio formats

• Fix message deserialization error with Argmax local server

• Split FAB into separate Join and Start listening buttons

  • When a meeting is detected, two buttons now appear: "Join (platform)" and "Start listening"
  • Join button opens the meeting URL without starting listening
  • Start listening button starts listening without opening a new tab
  • Join button hides on narrow screens to keep FAB inside note area

• Fix image attachment crashes and improve attachment handling

  • Images are now saved as files in session folder instead of base64, preventing app freezes
  • Preserve original attachment filenames (Obsidian-style)
  • Duplicate filenames automatically get numbered suffixes (e.g., "image 1.png")
  • Switch to official TipTap image extension with block-level images

• Make content folder migration safer

  • Content folder changes now copy data instead of moving it
  • Original folder is preserved for safety
  • Added confirmation dialog explaining the copy-only behavior

• Fix authentication token refresh errors

  • Handle refresh_token_already_used errors properly
  • Clear invalid sessions to prevent lock timeout warnings
  • Remove duplicate deep link handler

• Improve feedback modal

  • Add optional checkbox to attach redacted application logs
  • Logs include last 500 lines with user paths, emails, and IP addresses redacted

• Show spinner instead of checkmark while local STT is initializing

  • Prevents confusing UI state during model initialization

• Add sleep detection (macOS)

  • Recording automatically stops when you close your laptop lid or put the system to sleep
  • Prevents unintended recording while the system is inactive

• Enhanced context menu for calendar events

  • Events with notes now show: "Open in new tab", "Reveal in Finder", and "Delete completely"
  • "Delete completely" removes both the event and all associated session data
  • Ignoring events without notes now automatically cleans up empty sessions

• Fix credentials being lost after app restart

  • Session tokens now refresh automatically in the background
  • Switched to browser-native lock mechanism, eliminating spurious timeout warnings

• Fix memory growth during long recordings

  • Removed audio data accumulation that could consume ~19GB/day of memory
  • Added bounded audio channels with backpressure to prevent unbounded memory growth

• Fix Ollama model list not loading for remote instances

  • Model list now populates correctly when connecting to Ollama via Tailscale or other remote setups

• Fix starting recording from enhanced view

  • Clicking "Start listening" while viewing a summary now correctly switches to raw view

• Improved timeline event display

  • Events now use end time to determine past/future visibility
  • Event start times are displayed in the timeline sidebar

• Support custom base through CONTENT_BASE env and lab settings

• Fix changed tables detection in TinyBase persisters which might cause data loss

• Persist auth session across app updates

  • Stay logged in after updating Char

• Fix note title sync in left sidebar for event-attached sessions

  • Editing a note title attached to an event now updates the sidebar correctly

• Preserve ignored calendar events after app restart

  • Previously ignored events stay ignored after calendar sync

• Improve trial expiration handling

  • Show "Upgrade to Pro" for expired trials instead of "Start Free Trial"
  • Auto-reset cloud STT and Char LLM settings when trial expires

• Fix notification settings for mic detection

  • Mic detection notifications now respect the notification_detect setting

• Preserve 'me' label when merging contacts

  • The 'me' contact label is no longer lost when merging duplicate contacts

• Add git hash display in feedback modal

• Fix potential data loss from automatic stamp repair

  • Removed stamp repair logic that could incorrectly modify stored data
  • Improves data integrity during sync operations

• Fix transcript migration for existing sessions

  • Resolves migration issues with transcript.json files

• Fix date-fns panic errors in timeline and session metadata

  • Prevents invalid time value errors when displaying dates
  • Improves reliability of date parsing and formatting

• Optimize transcript word rendering and scrolling performance

  • Faster rendering of transcript words during active transcription
  • Smoother scrolling experience in transcript view

• Improve audio visualizer amplitude computation

  • Better perceived loudness using RMS (Root Mean Square) instead of peak detection
  • Logarithmic scaling that matches human hearing perception
  • Smoother animations with reduced jitter
  • Noise gate to hide background noise floor

• Websocket connection timeout tuning

• Add dynamic recording indicator overlay on dock icon

  • Visual indicator shows when audio recording is active
  • Recording state persists when dock icon changes during active recording

• Add icon buttons to show model in Finder and delete model in AI settings

  • Quick access to model files from settings
  • Delete models directly from AI settings interface

• Add in-app feedback modal for bug reports and feature requests

  • Integrated feedback system for user reports
  • Streamlined bug reporting workflow

• Attempt to fix invalid time value errors in startup.

AI & Model Selection

• Add ElevenLabs STT adapter support

  • Use ElevenLabs for speech-to-text transcription

• Improve Deepgram language variant recommendations

  • Better region-specific language variant suggestions for Deepgram adapter

• Add language normalization in adapters

  • Deduplicate languages with same ISO639 code
  • Prefer base language variants over regional variants when both are present

• Fix hardcoded Char path in language configuration

Audio

• Fix OGG audio generation and enable stereo output

  • Improved audio quality with proper stereo encoding
  • Fixed OGG file generation issues

• Fix OpenAI proxy to use correct sample rate and add intent=transcription parameter

  • Ensures proper audio format for OpenAI transcription API

Transcript

• Prevent partial words from leaking to inactive session transcripts

  • Fixes issue where incomplete words could appear in transcripts of inactive sessions

• Prevent active listening session tab from being replaced

  • Active transcription sessions are now protected from accidental tab replacement

Data & Storage

• Fix session loading to search recursively in subfolders

  • Sessions in nested folders are now properly discovered and loaded

• Fix TinyBase stamp leaking issue

  • Prevents data corruption by properly flushing and syncing after local persister load

• Fix template_id field in imported enhanced notes to prevent duplicates

  • Ensures proper template association when importing notes from previous versions

UI

• Fix timeline bucket sorting order in left sidebar

  • Week and month buckets now display in correct chronological order
  • Future events properly sorted with weeks before months

• Add pinned contacts with reordering support

  • Pin frequently used contacts for quick access
  • Drag and drop to reorder pinned contacts

• Add Report Issue button to error page

  • Includes device information for easier troubleshooting

• Fix conditional styling for skip reason text color

  • Skip reason banners now display with appropriate text color based on context

Calendar

• Fix recurring event sync and recurrence series ID population
  • Recurring events now properly sync and maintain their series relationships

Plugins

• Add AWS Bedrock plugin
  • List foundation models from AWS Bedrock
  • Filter models by provider, customization type, output modality, and inference type
  • Fix async AWS config loading to prevent blocking issues

Settings

• Move base and obsidian_vaults from path2 to settings plugin
  • Consolidates path-related settings into unified settings plugin

Developer Experience

• Refactor LLM proxy with provider abstraction

  • Extract OpenRouter-specific code into provider abstraction
  • Add e2e tests and CI workflow for LLM proxy

• Refactor transcribe proxy

  • Add tests and reduce code duplication in relay module
  • Improve code maintainability

• Move scroll-fade imports to @hypr/ui package

  • Better code organization and reusability

• Refactor hooks: rename use-permissions to usePermissions

  • Consistent naming convention across codebase

Web

• Add content admin UI
  • Manage website content including articles, changelog, and documentation
  • Browse and edit MDX files with integrated editor

Transcript

• Add regenerate transcript feature
  • Regenerate transcriptions from existing audio files
  • Accessible from the note header

Audio

• Improve macOS speaker capture performance

  • Make speaker capture real-time safe
  • Reuse conversion buffer for better efficiency

• Filter out internal tap devices from audio device selection

  • Prevents internal system devices from appearing in user-facing device lists

AI & Model Selection

• Improve model selection when using Char cloud

  • Better automatic model selection based on language and provider compatibility

• Enhance language support checking for STT providers

  • Improved language compatibility evaluation for Deepgram and AssemblyAI adapters
  • Better language configuration handling across adapters

UI

• Add gap between audio player and transcript for better visual separation

AI & LLM

• Add Mistral support for LLM provider

Speech-to-Text

• Add language support checking for STT providers
  • Check if selected provider/model supports configured languages
  • Suggest compatible providers based on selected languages
  • Show warnings when language/provider combination is not supported

Settings

• Apply system language defaults on settings load

UI

• Move chat trigger up when audio timeline is visible

Fixes

• Fix timeline date parsing to use local timezone instead of UTC
• Fix importer failing when encountering null words

Data & Storage

• Add migration for transcript file naming (_transcript.json → transcript.json)

Audio & Settings

• Add audio priority selector in settings
• Add audio section in settings
• Fix listen mode heuristic around headphones with wrong terminal type
• Increase speaker buffer size for better audio quality
• Fix missing permission for detect plugin

Data & Storage

• Replaced sqlite-based persisters with json and markdown-based persisters
• Avoid accidental data loss on load failure
• UUID folder migration for better data organization

UI & Navigation

• Expandable fields in metadata panel
• Swap Inter for SF Pro fonts
• Add "Open in new tab" context menu option
• Force pin listening tab with fixed left section
• Scroll left sidebar to 'now' indicator on initial load
• Add 'coming soon' tooltips to unimplemented features
• Show chat coming soon indicator

AI

• Add display name for Char auto model
• Add toast notification for logged-out users with Pro models configured
• Enable dismissing AI onboarding toast
• Add congratulatory modal when pro trial begins

Authentication

• Preserve auth session by saving store on app exit
• Enhance authentication flow with redirect handling

Calendar

• Parse date strings as local time to fix timezone offset

Performance

• Optimize scroll fade performance and memoize overlay
• Improve title input with local state management
• Reduce delay for toast appearance
• Move amplitude observation to ingest time for faster visualization

Char 1.0.1 brings significant improvements to plugin architecture, data management, and overall user experience.

Calendar Integration

• Display calendar event info in metadata chip
  • Show video icon and truncated event title
  • Display event details including title, location, meeting link, and date/time
  • Add Join button for meeting links

Auto-Enhancement & Title Generation

• Automatically switch to summary tab and start summarizing when listening stops
• Re-trigger summarization even if summary already exists
• Require minimum words before auto-enhancement triggers
• Show dismissible consent banner when listening starts (fades out after 5 seconds)
• Display skip reason banner when auto-enhance is skipped
• Prevent unnecessary title regeneration when switching tabs
• Restore focus to memos editor when switching tabs during meeting

Plugins

• Consolidate export, frontmatter, and folder plugins into unified fs-sync plugin

  • Move folder operations to fs-sync plugin
  • Move export functionality to fs-sync plugin
  • Move frontmatter operations to fs-sync plugin
  • Add audio operations (delete, exist, import, path) to fs-sync plugin
  • Add chat directory, entity directory, and scan utilities to fs-sync plugin
  • Add orphan cleanup commands (cleanup_orphan_dirs, cleanup_orphan_files)

• Add apple-contacts plugin

  • Implement Apple Contacts integration for importing contacts
  • Add import command for batch importing contacts

Persistence & Storage

• Major persister consolidation and refactoring
  • Merge note and transcript persisters into single session persister
  • Add chat shortcuts persister for storing chat shortcuts
  • Remove memories feature and related code
  • Refactor persister utilities into shared factories
  • Add factory helpers for JSON file and markdown directory persistence
  • Add generic transforms in human persister
  • Add values persister for key-value pairs
• Store transcript words inline instead of separate table
• Store array fields as JSON instead of comma-separated strings
• Move language settings to top-level 'language' key
• Remove extension_state and ai_providers from main store

Search & Indexing

• Tantivy search engine improvements
  • Refactor to consolidate search args into SearchRequest struct
  • Phase 1 foundation fixes for better search infrastructure
  • Phase 2 feature parity with Orama search capabilities
  • Add advanced search features including filters and sorting
  • Improve batching, concurrency, and index versioning

Audio & Devices

• Add DeviceListChanged event for device connect/disconnect detection
• Fix wrong sample rate returned from audio buffer
• Add com.apple.Sound-Settings.extension to microphone detection exclusion list
• Add error propagation from STT adapters
  • Improve error handling in Argmax, AssemblyAI, Fireworks, Gladia, OpenAI, and Soniox adapters
  • Better error reporting in listener actor

Speech-to-Text

• Fix Deepgram language configuration
• Remove unsupported detect_language option from Deepgram streaming
• Don't force accordion in STT config UI

Editor

• Add resizable image extension to Tiptap editor
  • Resize handles with hover effects
  • Preserve custom attachmentId attribute

Data Import

• Add Templates import to Char importer (supports v0 stable and nightly)
• Refactor importer plugin with simplified import flow
• Remove LIMIT in session queries for better import handling
• Remove auto import on app startup
• Remove data tab from main navigation (consolidated into settings)
• Update data import UI in settings
• Fix path display and reveal behavior for Char sources
• Improve legacy HTML-to-markdown conversion
• Use Tiptap native markdown conversion for better parsing

Performance

• Implement incremental persister save optimization using content hashing
• Optimize TinyBase store by inlining nested words

Meeting View

• Prefer listener view when re-entering a meeting

macOS Notification

• Improve progress bar animation with correct pause/resume behavior
• Move progress bar to bottom of notification

Fixes

• Fix edge-case for is_builtin_display_inactive on macOS
• Compute Ollama Origin header from baseUrl to fix Forbidden error
• Simplified error UI with single "Restart App" button
• Read AI language from settings store instead of main store
• Normalize bullet points (•) to dashes (-) in AI summaries
• Better Sentry config with user-id tracking
• Improve accessibility for tab button interaction

Calendar Integration

• Display calendar event info in metadata chip
  • Show video icon and truncated event title
  • Display event details including title, location, meeting link, and date/time
  • Add Join button for meeting links

Auto-Enhancement

• Automatically switch to summary tab and start summarizing when listening stops
• Re-trigger summarization even if summary already exists

Audio & Devices

• Add DeviceListChanged event for device connect/disconnect detection
• Fix wrong sample rate returned from audio buffer
• Add com.apple.Sound-Settings.extension to microphone detection exclusion list

Speech-to-Text

• Fix Deepgram language configuration
• Remove unsupported detect_language option from Deepgram streaming
• Don't force accordion in STT config UI

Data Import

• Improve legacy HTML-to-markdown conversion
• Use Tiptap native markdown conversion for better parsing

Fixes

• Read AI language from settings store instead of main store
• Normalize bullet points (•) to dashes (-) in AI summaries
• Better Sentry config with user-id tracking

Meeting View

• Prefer listener view when re-entering a meeting

Data Import

• Fix path display and reveal behavior for Char sources
• Store transcript words inline instead of separate table

macOS Notification

• Improve progress bar animation with correct pause/resume behavior
• Move progress bar to bottom of notification

Auto-Enhancement & Title Generation

• Require minimum words before auto-enhancement triggers
• Show dismissible consent banner when listening starts (fades out after 5 seconds)
• Display skip reason banner when auto-enhance is skipped
• Prevent unnecessary title regeneration when switching tabs
• Restore focus to memos editor when switching tabs during meeting

Editor

• Add resizable image extension to Tiptap editor
  • Resize handles with hover effects
  • Preserve custom attachmentId attribute

Data Import

• Add Templates import to Char importer (supports v0 stable and nightly)
• Refactor importer plugin with simplified import flow
• Remove LIMIT in session queries for better import handling
• Remove auto import on app startup
• Remove data tab from main navigation (consolidated into settings)
• Update data import UI in settings

Performance

• Implement incremental persister save optimization using content hashing
• Optimize TinyBase store by inlining nested words
• Improve Tantivy search batching, concurrency, and index versioning

Fixes

• Compute Ollama Origin header from baseUrl to fix Forbidden error
• Simplified error UI with single "Restart App" button

Other

• Store array fields as JSON instead of comma-separated strings
• Move language settings to top-level 'language' key
• Add values persister for key-value pairs
• Remove extension_state and ai_providers from main store
• Improve accessibility for tab button interaction

Plugins

• Consolidate export, frontmatter, and folder plugins into unified fs-sync plugin

  • Move folder operations to fs-sync plugin
  • Move export functionality to fs-sync plugin
  • Move frontmatter operations to fs-sync plugin
  • Add audio operations (delete, exist, import, path) to fs-sync plugin
  • Add chat directory, entity directory, and scan utilities to fs-sync plugin
  • Add orphan cleanup commands (cleanup_orphan_dirs, cleanup_orphan_files)

• Add apple-contacts plugin

  • Implement Apple Contacts integration for importing contacts
  • Add import command for batch importing contacts

Persistence & Storage

• Major persister consolidation and refactoring
  • Merge note and transcript persisters into single session persister
  • Merge persister/note into persister/session
  • Add chat shortcuts persister for storing chat shortcuts
  • Remove memories feature and related code
  • Refactor persister utilities into shared factories
  • Add factory helpers for JSON file and markdown directory persistence
  • Move frontmatter loading to fs-sync plugin
  • Add generic transforms in human persister
  • Move scan utils into fs-sync plugin

Search & Indexing

• Tantivy search engine improvements
  • Refactor to consolidate search args into SearchRequest struct
  • Phase 1 foundation fixes for better search infrastructure
  • Phase 2 feature parity with Orama search capabilities
  • Add advanced search features including filters and sorting

Audio & Transcription

• Add error propagation from STT adapters
  • Improve error handling in Argmax, AssemblyAI, Fireworks, Gladia, OpenAI, and Soniox adapters
  • Better error reporting in listener actor

Bug Fixes

• Fix edge-case for is_builtin_display_inactive on macOS

Developer Experience

• Add changelog workflow for automated changelog generation
• Add clean-plugins utils for plugin management

We're excited to announce Char 1.0.0, our first stable release! This release consolidates all the improvements from our nightly builds into a polished, production-ready experience.

Platform Support

• macOS support for Apple Silicon (aarch64) and Intel (x86_64)
• Linux support with AppImage and .deb distributions
• Standardized artifact naming with architecture info

Speech-to-Text Providers

• Deepgram - Full support with API key validation and keyword handling
• Gladia - Complete support with code switching, partial transcripts, multi-channel audio, custom vocabulary, and speaker diarization
• OpenAI - Realtime and batch transcription
• AssemblyAI - Streaming and batch transcription
• Soniox - STT provider support
• Whisper Large V3 - Local STT option
• Language detection and multi-language support across providers

AI & LLM Support

• Google Generative AI provider
• OpenAI and Anthropic model providers
• Ollama support with CORS fix
• LLM configuration error rendering in editor

Audio & Recording

• Scrolling waveform visualization in the listen button
• Fixed recording issues with headphones (sample rate, distortion)
• VAD (Voice Activity Detection) improvements with continuous masking
• AGC (Automatic Gain Control) with VAD integration
• Dual-websocket mode for multichannel support with local STT
• 24k sample-rate on macOS to minimize resampling
• Linux audio support via PulseAudio with headphone detection

Search & Indexing

• Tantivy-based full-text search replacing Pagefind
• Language detection and language-aware tokenization
• Multi-collection indexing for scalable search
• Search results with highlighted matches

Folder Management

• Create, rename, delete, and move sessions between folders
• Reveal session in Finder
• Folder detection based on directory structure

Export

• PDF export with typst-based formatting
• JSON export with batch support
• VTT export with speaker information
• Transcript export improvements
• Tab-aware export menu (memo, enhanced, transcript)

Editor

• Tiptap document JSON validation
• List navigation and backspace behavior improvements
• Tab and shift-tab support for list indentation
• Prompt editor with readonly ranges support
• 20x faster tiptap JSON to markdown export with custom Rust implementation

Chat

• Dedicated chat editor with slash command support
• Chat trigger on hover near bottom
• Resizable chat window

Settings

• Unified settings.json configuration
• Prompt, shortcut, template, and extensions tabs
• Custom template support
• App settings organized into tabs

Window & UI

• Window size based on screen size
• Persistent window position and size across restarts
• macOS traffic light button positioning
• Native menu support
• Keyboard shortcuts for tab navigation
• Tab restore functionality
• Progressive blur effect at scroll ends
• Responsive search with manual expansion
• 20% rendering optimization for long transcripts
• Fancy NotFound and ErrorBoundary components

Onboarding

• Redesigned onboarding experience with background music
• Free trial support
• Auto model and provider selection for Pro trial
• Back navigation in permissions and configuration screens

Updates & Distribution

• OTA (Over-the-Air) updates with cache support
• Tray menu with update check, status dialogs, and settings
• Changelog tab shown automatically after updates
• CLI update command
• Homebrew distribution

Notifications

• macOS notification enhancements
• Notification collapse animation
• Auto-start trigger improvements

Plugins

• JS Plugin - JavaScript code evaluation via rquickjs
• Notify Plugin - Filesystem watching with debouncing
• Overlay Plugin - Floating controls with click-through behavior
• Hooks Plugin - Event handling with 5-second timeout for user scripts
• Listener2 Plugin - Batch and subtitle functionality
• Updater2 Plugin - Update check functionality
• Deeplink2 Plugin - Type-safe deep link handling
• Network Plugin - Network availability monitoring

Calendar

• Calendar picker UI with permission handling
• Contact view opens in new tab when clicking participant chip

CLI

• Experimental CLI support
• CLI installer in app menu
• --help and --version flags

Developer Experience

• Deno-powered extensions runtime with iframe-based isolation
• Separate devtool window
• E2E tests using Linux WebDriver
• PostHog analytics fixes
• Sentry monitoring integration

Other Improvements

• Prevent closing tab or quitting app during active listening
• Session failure error surfacing in listen button tooltip
• Disable listen button when offline with non-local model
• Title generation after meetings
• Inline codeblock wrapping
• Sound effects playback fixes
• Keyboard navigation improvements

Search & Indexing

• Replace Pagefind plugin with Tantivy-based full-text search plugin
  • Implement Tantivy search engine as Orama replacement for better performance and scalability
  • Add language detection and language-aware tokenization for improved search accuracy
  • Support multi-collection indexing for scalable search across different document types
  • Merge search and search_fuzzy into single unified search function with options
  • Simplify API to expose only search and reindex commands
  • Refactor tantivy and language crates for better organization
  • Improve orphan removal error handling in search index

Folder Management

• Add new folder plugin for organizing sessions
  • Implement folder operations: create, rename, delete, and move sessions between folders
  • Add unified folder operations layer for bidirectional persistence
  • Remove _default folder concept - sessions without folders now live directly under sessions/
  • Add folder migration persister to handle legacy folder structure
  • Implement folder detection based on directory structure (folders don't contain _meta.json)
  • Add reveal session in finder command

Persistence & Storage

• Major persister refactoring
  • Refactor with createSessionDirPersister helper for consistent session directory handling
  • Add createSingleTablePersister helper for simpler single-table persistence
  • Migrate existing persisters to use new helper functions
  • Refactor local persister, folder persister, chat persister, setting persister
  • Refactor calendar and event persisters with store merging
  • Remove PersisterMode enum
  • Remove legacy enhanced_md from session store
  • Add per-path content hashing to reduce unnecessary file writes
  • Fix folder persistence issues and add missing save-only persisters

Plugins

• Add JS plugin with eval command using rquickjs

  • Enable JavaScript code evaluation from Rust backend
  • Useful for scripting and automation

• Add notify plugin for filesystem watching

  • Implement filesystem change notifications with debouncing
  • Add start/stop commands for watching file changes
  • Emit FileChanged events when files are modified
  • Add default permission set for notify plugin

• Add overlay plugin extracted from windows plugin

  • Implement overlay window functionality for floating controls
  • Add fake window bounds tracking for click-through behavior
  • Support overlay listeners with configurable options

• Remove pagefind plugin (replaced by tantivy)

Export

• Add JSON export support in export plugin
  • Support exporting sessions and data as JSON format
  • Add exportJsonBatch command for batch JSON exports
  • Make overflow menu export tab-aware (memo, enhanced, transcript)

UI & UX Improvements

• Chat UI design cleanup

  • Polish empty view appearance
  • Show chat trigger when hovering near bottom
  • Improve chat input UI and interactions
  • Update chat window minimum width
  • Resize window on right chat toggle

• Profile menu improvements

  • Polish expandable profile menu
  • Remove keyboard shortcuts from profile menu items

• Settings improvements

  • Add tabs to app settings tab
  • Add trial expiration modal in dev panel

• Notification improvements

  • Refactor notification-macos with rustbridge
  • Reorganize Swift file structure for better maintainability
  • Add macOS notification enhancement with serialized arg passing
  • Adjust close button in notification and add timer
  • Fix notification collapse animation
  • Fix notification UI and auto-start trigger
  • Make path relative in notify event

• Title generation

  • Add title generation button
  • Improve auto-title generation
  • Update title generation prompt
  • Generate title after meeting

• Other UI improvements

  • Fix inline codeblocks to wrap instead of overflow
  • Surface session failure error in listen button tooltip
  • Disable listen button when offline with non-local model
  • Add root supervisor in listener for better session failure handling

Audio & Transcription

• Note enhancing changes
• Various refactors in audio actor system

Control Window

• Add overlay plugin architecture improvements
• Add Control window UI for quick access to recording controls
• Add widget expand/collapse functionality to control window
• Various control window visibility groundwork

Developer Experience

• Write AGENTS.md to base directory on app startup
• Add permission check troubleshooting utils in frontend and dev
• Preserve inner tab scroll; skip transcript restore during meetings

Other Improvements

• Improve orphan removal error handling
• Various refactors and code cleanup
• Fix tests and improve error handling

• Fix file upload and batch transcription.
• Various refactors in audio actor system.
• Fix outdated navigation event from tray icon.
• Fix session deletion logic.
• Improve WebSocket error handling. E.g. no retry on auth error.
• Keyboard navigation improvements.
• Other UI improvements and refactors.

• Fix search results to display proper UI with highlighted matches instead of raw JSON.
• Hide floating action buttons (FABs) when typing near the bottom of the editor.
• Fix native menu item panic and action handling.
• Fix most of the problems existing in the onboarding flow.
• Enable the open -a "Char Staging" --args --onboarding $(date +%s) command for reliable QA in staging environment.

• Fix race condition in Windows command operations.
• Refine template rendering, including proper output language instructions.

• Fix onboarding freeze issue.
• Separate data folder for staging environment.

• Add calendar picker UI with permission handling.
• Improve templates tab view.
• Add draggable header region to error and 404 screens.
• Add scroll fade overlay for better scrolling UX.
• Open contacts view in new tab when clicking participant chip.
• Add speaker information to VTT export.
• Fix traffic light position.
• Fix sound effects playback issues.
• Migrate to TinyBase 7.3.

• Fix fatal error around keywords handling for Deepgram and Gladia.
• Optimize tiptap json to markdown export performance by 20x with custom rust implementation.
• Await beforeListeningStarted hook before starting session.
• Add 5 second timeout for user-provided hook scripts.

• New onboarding experience with background music.
• Add PDF export with typst-based formatting.
• Fix transcript export.
• Add tab restore functionality.
• Add keyboard shortcuts for tab navigation and new tabs.
• Show keyboard shortcut badges on sidebar menu items.
• Add progressive blur effect at scroll ends in AI view.
• Use native menu.
• Fix spoken language configuration for live STT.
• Migrate from Typescript to Rust for Char Cloud Proxy.

• Fix OTA update check interval.

• Remove react-scan that was accidentally included in the production build.
• Add OTA updates installed from cache.

Session

• Prevent closing tab during active listening.
• Prevent quitting app during live session.
• Add Deepgram API key validation.

Window

• Fix window sizing validation.
• Add min/max size caps to prevent oversized windows.
• Fix traffic light position for macOS.

Others

• Rendering optimizations for long transcript. (around 20% improvement)
• Responsive search with manual expansion.
• Various UI fixes and refactors.

Notification

• Fix handlers for notification events.

Window

• Window size is now decided based on the screen size.
• Window position and size are now persisted across restarts.
• Traffic lights are now positioned correctly across different macOS versions.

Others

• Mostly UI fixes and refactors.

Onboarding

• Auto set model and provider when Pro trial is enabled during onboarding.

Others

• Fix auto-update flow.
• Fix deep-link event registration.
• Fix small UI issues.

Onboarding

• Redesign onboarding UI.
• Add back navigation to permissions and configuration screens.
• Fix onboarding login flow.

Editor

• Improve list navigation and backspace behavior.
• Add tab and shift-tab support for list indentation.

Others

• Add sub-menus to settings menu in tray.

OTA

• Extract update check functionality into a new updater2 plugin.
• Add "Check for Updates" and "Settings" menu items to the tray with icons and status dialogs.
• Add update command to CLI.
• Add new updater behaviors and UI, including a 3-state conditional update menu in the tray.
• Show changelog tab automatically after app update.

Settings

• Replace Settings window with multiple dedicated tabs for better organization and user experience.
• Implement unified settings.json by merging hooks.json into settings.json with top-level merging.

Others

• Fixed Homebrew distribution.
• Conditionally hide tab chat button based on tab type.
• Add refresh button to model combobox when list is empty.
• Improve transcript auto-scroll behavior when user scrolls up.
• Fix handling of stale sessions on app launch and during sign out.
• Remove email login functionality.
• Add Lithuanian language support.
• Add Pro icon and runtime dock icon modification.
• Adjust traffic light position for better macOS window controls alignment.
• Update ONNX runtime to latest version.
• Add autocorrect=off attribute to disable browser autocomplete in editors.
• Various UI fixes and improvements.

• Add app info to MicStopped event, so that the app ignored list can also be applied to MicStopped events, consistent with MicStarted events.
• Fix store data loading from disk in non-main windows (e.g., Settings), eliminating a 1-2 second delay.
• Fix analytics app_version to use the APP_VERSION environment variable.
• Add scrolling waveform visualization showing audio amplitude in the listen button.

• Add Export Transcript menu item to session overflow menu.
• Extract device-heuristic crate with improved channel mode detection for audio devices.
• Remove unused calendar crates (outlook, google) and inline apple-calendar crate.
• Remove large unused dependencies (kalosm, graph-rs-sdk, apalis-sql) for smaller builds.
• Fix PostHog analytics (stale API key, empty version)

• Implement free trial support in desktop app with new onboarding flow.
• Add full Gladia STT support with code switching, partial transcripts, multi-channel audio, custom vocabulary, and speaker diarization.
• Fix hooks plugin functionality and improve event handling.
• Introduce keys.json for better state management in TinyBase store.
• Improve profile name rendering in sidebar.

• Fix Ollama CORS error by adding Origin header for requests.
• Use different deeplink schema per bundle for proper app routing.
• Apply clippy fixes to Rust crates (whisper-local, analytics, vvad, buffer, llama).
• Fix onboarding flow issues.

• Unify VAD logic with StreamingVad for better audio processing.
• Fix batch audio processing bugs in listener2 plugin.
• Update onboarding flow with new model selection.
• Fix release name for Sentry monitoring.

• Add Gladia STT provider with realtime and batch transcription support.
• Add OpenAI STT provider with realtime and batch transcription support.
• Add prompt, shortcut, template, and extensions tabs in settings.
• Add custom template support to template plugin.
• Update DMG background image for macOS installer.
• Small UI fixes.

• Add .deb build for Linux x86_64.
• Add network availability monitoring plugin.
• Add NetworkContext for network status in desktop app.
• Add dedicated chat editor with slash command support.
• Add AssemblyAI adapter for streaming and batch transcription.
• Add main language configuration support.
• Add fancy NotFound and ErrorBoundary components with animations.
• Skip permissions step in onboarding on Linux.
• Improve language support in Deepgram-compatible adapters.
• Refactor STT adapter selection.
• Fix model download issues.
• Small Linux fixes.

• Add Intel macOS (x86_64) support.
• Add Deno-powered extensions runtime with iframe-based isolation and TinyBase sync.
• Add calendar extension. (not included yet)
• Add VTT export functionality to listener2 plugin.
• Add macOS-style traffic light buttons for Linux.
• Add Whisper Large V3 support to STT settings.
• Add Soniox STT provider.
• Remove native titlebar on Linux for custom titlebar support.
• Improve STT health check status in listen button.
• Improve language handling per STT adapter.
• Fix Linux notification window issue.
• Fix deeplink for Linux.
• Fix infinite loop in TranscriptContainer selectors.

• Remove memory settings from preferences.
• Fix macOS audio tap.
• Fix CLI --help and --version flags handling.
• Add listener2 plugin with batch and subtitle functionality.

• Fix AGC with VAD and config modification.
• Add Linux implementation for app listing features and device monitoring using PulseAudio.
• Add Linux headphone detection support for automatic channel mode switching.
• Dynamic linking ONNX runtime for improved performance.
• Replace alphabetic speaker labels (A, B, C) with numeric indexing (1, 2, 3) for unknown speakers.
• Add streaming state to VadAgc for cross-call framing.
• Ensure Summary tab always renders when transcript exists.
• Add dual-websocket mode as multichannel workaround for local STT providers.
• Add prompt editor with readonly ranges support.
• Improve Linux system audio permission check.
• Fix audio autoplay issues by adding URL guard in AudioPlayerProvider.
• Implement continuous VAD masking for microphone stream.
• Add deeplink2 plugin for type-safe deep link handling.
• Fix shutdown handling during realtime audio and await control sends.
• Fix duplicated enhanced tab and enhance streaming UI.
• Add amplitude floor in VAD masking.
• Add root supervisor in desktop app.
• Enhance transcript store with partial hints preservation and channel-aware speaker hints.

• Add "Show in Finder" option to overflow menu.
• Fix recording issues when using headphones. (Wrong sample rate, Distorted recording)
• Add CLI installer in app menu.
• Fix sample drop in resampler on rate change boundary.
• Add basic E2E tests using Linux WebDriver.
• Initial Linux support with AppImage distribution. (Not functional yet)
• Standardize artifact naming with architecture info (hyprnote-macos-aarch64.dmg, hyprnote-linux-x86_64.AppImage).

• 1.0.0-nightly.5 add experimental CLI, which breaks the app in release builds. This release fixes the issue.

• Show only version string for OTA status.
• Use 24k sample-rate in macOS to minimize resampling.
• Prevent drag-and-drop by default.
• Add experimental CLI support.

• Add Google Generative AI provider support.
• Fix OpenAI and Anthropic model provider support.
• Add separate window for devtool.
• Add LLM config error rendered in editor area.
• Introduce hooks plugin.

• Validate Tiptap document JSON before initializing editor.

• Fix app icon padding.

• Initial release.