Add Dependabot config for automated dependency updates#24
Merged
Conversation
Default setup used build-mode 'none' for C# which resulted in low analysis quality (call target resolution 83% vs 85% threshold). Switching to autobuild resolves NuGet dependencies properly. Other languages (JS/TS, Python, Java) remain on build-mode 'none'. Upgraded to security-extended query suite. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Covers 28 directories across 4 ecosystems: - pip (5 Python dirs) - npm (10 JS/TS dirs including @azure/ai-voicelive) - nuget (6 C# dirs) - maven (2 Java dirs) Weekly schedule, voicelive-repo-maintainers as reviewers, language-specific labels for filtering. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
goergenj
added a commit
that referenced
this pull request
Mar 25, 2026
* Switch CodeQL to advanced setup with autobuild for C# Default setup used build-mode 'none' for C# which resulted in low analysis quality (call target resolution 83% vs 85% threshold). Switching to autobuild resolves NuGet dependencies properly. Other languages (JS/TS, Python, Java) remain on build-mode 'none'. Upgraded to security-extended query suite. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Add Dependabot config for all package directories Covers 28 directories across 4 ecosystems: - pip (5 Python dirs) - npm (10 JS/TS dirs including @azure/ai-voicelive) - nuget (6 C# dirs) - maven (2 Java dirs) Weekly schedule, voicelive-repo-maintainers as reviewers, language-specific labels for filtering. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
goergenj
added a commit
that referenced
this pull request
Mar 25, 2026
* Phase 1: Add TopBar, remove StartScreen, auto-connect on load - Add TopBar component with agent name, Voice/Text segmented toggle, New Thread button, and settings gear icon - Add useUrlParams hook for ?mode=voice|text and ?lock=true params - ?mode= locks input mode and hides toggle - ?lock=true hides settings gear + toggle entirely - Remove StartScreen — app auto-connects WebSocket on mount - Remove SessionEnded screen — idle state shows agent details - Settings panel now accessible from TopBar gear icon in all states - Frontend builds and type-checks cleanly Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Phase 2: Voice mode polish — fix layout, remove dead components - Fix ActiveSession height from 100vh to 100% (works with TopBar) - Delete StartScreen.tsx and SessionEnded.tsx (no longer used) - Session end now returns to idle state with orb in content area Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Phase 3: Text mode via Voice Live WebSocket - Add ChatMessages component (scrollable bubbles, user right/assistant left) - Add ChatInput component (text bar + send button, Enter to send) - Add sendTextMessage to useVoiceSession hook (sends 'send_text' WS msg) - Add send_text handler to Python voice_handler (conversation.item.create + response.create via Voice Live SDK) - Wire send_text in Python app.py WebSocket message router - TopBar Voice/Text toggle switches between VoiceOrb and Chat views - Fix auto-connect race: gate on configLoaded flag (review feedback) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Phase 4: Agent details, starter messages, waves background - Add AgentPrimaryDetails component (icon, name, description) - Add StarterMessages component (clickable prompt cards) - Add Waves component (animated SVG background for idle state) - Replace basic idle VoiceOrb with rich agent details + prompts - Waves animate behind idle content, pause when session is active Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix review findings: session leak, mic capture, transcript duplication - Fix New Thread: call stopSession before resetSession to close WS/capture - Fix text mode: skip mic capture when inputMode is 'text' via ref - Fix text mode: add end-session button alongside ChatInput - Fix ChatMessages: filter out non-final streaming duplicates - Sync inputModeRef from App to useVoiceSession hook Addresses all 3 high-severity findings from Codex 5.3 review. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Phase 5: Visual polish — update BuiltWithBadge to Microsoft Foundry - Update badge text from 'Azure AI Foundry' to 'Microsoft Foundry' matching the Foundry Portal's branding Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix final review: starter prompts, setInputModeRef stability - Fix starter prompts: start session if idle before sending text, queues message to fire after connection establishes - Wrap setInputModeRef in useCallback to prevent useEffect re-runs on every render (new function identity issue) Addresses Opus 4.6 final review findings. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Add Playwright smoke tests for new frontend UX 9 tests covering: TopBar controls, Voice/Text toggle, URL params (?lock=true, ?mode=text), settings panel, built-with badge, mode switching. All pass against local Python backend. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Visual overhaul: match Foundry Portal design exactly TopBar: - Replace segmented toggle with ··· dropdown menu - Menu has Settings, Terms of use, Privacy, Send feedback - New chat button with text+icon label Idle state: - Re-introduce Start Session flow (agent heading + description + solid purple orb + 'Lets talk' + Start session pill button) - Gray waves background matching Foundry screenshot Active voice orb: - 3 concentric circles (244/182/138px ratio) with graduated opacity - Pulse animation 1.8s for listening state - Solid circle in idle (120px) Theme: - Default to light mode (was dark) - Gray waves (was purple) URL params: - Added ?agent=, ?project= (pre-fill agent config) - Added ?theme=light|dark (override theme) - Added ?greeting=false (suppress proactive greeting) Badge: - Updated to 'Build & deploy AI agents with Microsoft Foundry →' Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix review findings: connecting overlap, locked mode, variable ordering, tests - Fix text mode connecting overlap: add state !== 'connecting' guard - Fix locked mode dead-end: don't disable Start when isLocked - Fix URL params: only force agent mode when BOTH agent+project provided - Fix variable ordering: move isActive/isIdle before handleNewThread - Rewrite Playwright tests for new TopBar (··· menu, Settings from menu) - All 8 Playwright tests pass Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fine-tune visual details to match Foundry Portal screenshots - Theme: change from segmented buttons to dropdown in Settings - TopBar: use bg-1 background (lighter), round ··· button, thinner icons (1.5 strokeWidth) - BuiltWithBadge: two-row layout matching screenshot (line 1: small text, line 2: bold brand + arrow) - Menu icons: 18px with 1.5 stroke for cleaner Fluent-like appearance Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Align font sizes to match Foundry Portal exactly All sizes now use px values matching the Foundry CSS: - Agent heading: 20px (was 1.6rem/26px) - Description: 14px (was 0.95rem/15px) - Let's talk: 20px (was 1.25rem/20px - already correct) - Talk description: 14px, max-width 280px - Start button: 14px (was 1rem/16px) - TopBar agent name: 14px (was 0.95rem) - New chat button: 14px, weight 400 - Menu items: 14px - Badge: 12px/14px Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Align spacing and sizing to Foundry design tokens Mapped to Fluent UI spacing tokens (spacingS=8, spacingL=16, spacingXL=20, spacingXXL=24): - Idle content gap: 8px (was 12px) matching labelBlock gap - Let's talk margin-top: 8px (was 16px) matching idleLabelSpacing - Start button: margin-top 24px (spacingXXL), font 16px, padding 8px 24px - Subtitle max-width: 250px (was 280px) matching subtitleText - Title lineHeight: 20px (was 24px) matching lineHeightBase300 - Controls: 40px buttons (was 48px), 8px gap (was 16px), 16px padding - Status text: 20px semibold (was 1.1rem/500) matching titleText Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Comprehensive UX alignment — fix all 33 gap analysis items P0 fixes: - VoiceOrb active circle ratios: 1.15x/1.52x/2.03x (138/182/244 from 120) - ActiveSession orb always 120px base (was 240/160) - Idle heading 28px (was 20px) for visual prominence - Start button min-width 200px, padding 8px 16px - Waves: 3 layers, --wave-color token, height 30% P1 fixes: - All secondary text: --fg-2 (was --fg-3) for colorNeutralForeground2 - Bubble borderRadius: 8px everywhere (was 12-16px) - Transcript text: 12px (was 15px), chat text: 14px - All rem values converted to px - Settings title: 20px/600 (was 1.3rem/700) - Settings labels: 14px/600/--fg-2 (was 0.85rem/500/--fg-3) - All input fonts: 14px, padding 8px 12px - CC label weight: 600 (was 700) - Mic button: purple accent ring (--voice-primary border) P2 fixes: - TopBar dots button: 40px (was 32px) - All SVG icons: 20px (was 16-18px) - All borderRadius on inputs/banners: 8px (was 10px) - Menu item gap: 8px, TopBar padding: 12px 20px - Send feedback icon: chat bubble (was person-group) P3 fixes: - --wave-color CSS token added (gray for light, subtle for dark) - Badge aria-label selector fixed - Optional badge: 12px (was 0.7rem) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix layout: heading at top, orb centered — match reference screenshot - Separate heading+description (top of content) from orb+labels+button (vertically centered) matching the Foundry Portal layout - Start button: auto-width (removed minWidth 200px), 14px font - Heading: 24px (was 28px), positioned at top not centered - TopBar: New chat grayed (--fg-3), dots button thinner border - Fix connecting state style reference Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Replace all icons with official Fluent UI 20px paths, lighten waves - TopBar: all 6 icons replaced with Fluent UI system icons (ChatAdd, MoreHorizontal, Settings, TextAlignLeft, Shield, PersonFeedback) - All SVGs now use fill='currentColor' (no stroke) matching Foundry - Waves: reduced opacity to 0.04/0.025 for subtler appearance - Wave color lightened (light: #e0e0e0, dark: rgba(255,255,255,0.05)) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Remove temp icon extraction script * Fix font, text colors, and wave visibility to match reference Fonts: - Add 'Segoe UI Variable' as primary font (Foundry Portal uses this) - Set base font-size 14px and line-height 20px on body Light theme colors: - bg-1: #fafafa (was #f5f5f5) — slightly warmer white - fg-2: #707070 (was #616161) — lighter secondary text - fg-3: #a0a0a0 (was #8a8a8a) — lighter tertiary text - wave-color: #d1cdc7 — warm taupe gray matching reference waves Waves: - Opacity 0.50/0.35/0.25 (was 0.04/0.025/0.04) — now clearly visible like the reference screenshot's layered gray waves Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Canvas waves, fix font colors and sizes to match Foundry exactly - Waves: rewritten from SVG to Canvas rendering matching Foundry code (3 layers, hsla colors, requestAnimationFrame, ResizeObserver) - Font colors: fg-2=#424242, fg-3=#616161 (matching colorNeutralForeground) - Agent heading: 20px (was 24px) matching Foundry titleText - Background: bg-1=#f5f5f5 (Fluent colorNeutralBackground3Selected) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Match Foundry VoiceOrb, controls, and session-ended exactly VoiceOrb — adapted from Foundry AudioPulse pattern: - Idle: 120px solid purple circle, no rings - Connected: two pulse rings (228px@0.2, 190px@0.3) behind core - Speaking: pulseMove animation (1.5s, scale 1.06) - Listening: pulseListening animation (1.8s, scale 1.06) - Sizes match Foundry (MIN_PULSE_SIZE=190, SCALE_FACTOR=1.2) SessionControls — adapted from Foundry actionBar: - Icon buttons: 40px, transparent bg, circular - Mic: branded border (--voice-primary), accent ring - Muted mic: filled bg, accent stroke - End session: text pill button (not X icon) SessionEndedView — new component: - Shows sessionID at top - Filtered final transcript entries - New chat button to restart Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix: align all frontend styles to Foundry design tokens - App.tsx: start button fontSize 14→16px, lineHeight 20→22px - TopBar.tsx: bar padding 20→16px, right gap 4→8px, --fg-3→--fg-2, rem→px - SettingsPanel.tsx: all --fg-3→--fg-2 (6 inline + 2 style objects), rem→px - TranscriptOverlay.tsx: bubble padding 12px 16px→8px 12px, role 0.75rem→12px, --fg-3→--fg-2 - ErrorBanner.tsx: close button 1rem→14px - BuiltWithBadge.tsx: logo 1.4rem→22px, line1 --fg-3→--fg-2 - AgentPrimaryDetails.tsx: 3rem→48px, 1.4rem→20px, 0.95rem→14px, --fg-3→--fg-2 - StarterMessages.tsx: 0.9rem→14px Zero rem values remain in components. Zero --fg-3 usage in components. tsc --noEmit and vite build pass clean. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix session-ended overlap, CC icon, purple color - Fix isIdle: exclude 'ended' state so SessionEndedView renders alone (was showing idle Start screen on top of ended transcript) - CC button: replace text 'CC' with Fluent ClosedCaption20Regular icon - Purple color: #7B5EA7 (was #7c5cbf) — matches Foundry colorBrandBackground Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * feat: adopt Fluent UI components across Voice Live frontend - Wrap app in FluentProvider with dynamic theme switching (light/dark) - Replace CSS theme vars with Fluent token aliases (with fallbacks) - App.tsx: Replace start button with Fluent Button - TopBar.tsx: Replace hand-built dropdown with Fluent Menu/MenuItem - SessionControls.tsx: Replace all buttons with Fluent Button components - SessionEndedView.tsx: Replace new chat button with Fluent Button - ActiveSession.tsx: Replace status text with Fluent Text component - ChatInput.tsx: Replace send button with Fluent Button + SendRegular icon - ErrorBanner.tsx: Replace close button with Fluent Button + DismissRegular - SettingsPanel.tsx: Replace all form controls with Fluent components (Select, Input, Textarea, Slider, Checkbox, Label, Text, Button) - Remove all manual menu state management (useState, useRef, useEffect) - Remove ~340 lines of hand-written style constants Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Adopt Fluent UI components + fix review findings Phase 1: Foundation - Add @fluentui/react-components dependency - Wrap app in FluentProvider with dynamic webLightTheme/webDarkTheme - Map all CSS vars to Fluent token aliases with fallbacks Phase 2: Component swaps - TopBar: Fluent Menu/MenuItem/Button (removed 100 lines of manual dropdown) - SessionControls: all buttons → Fluent Button with icons - App.tsx: Start session → Fluent Button appearance=primary shape=circular - SessionEndedView: New chat → Fluent Button - ActiveSession: status → Fluent Text - ChatInput: send → Fluent Button with SendRegular icon - ErrorBanner: close → Fluent Button with DismissRegular Phase 4: SettingsPanel - All form controls → Fluent Select/Input/Textarea/Slider/Checkbox/Label Review fixes: - Scope button hover/active CSS to non-Fluent buttons only - Remove !important from focus styles, scope to non-Fluent controls - Fix system theme detection: track via React state for FluentProvider re-render - Update Playwright tests for Fluent component selectors 8/8 Playwright tests pass. TypeScript + build clean. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix: use custom purple brand theme instead of default blue - Created theme.ts with purple BrandVariants ramp (base #7B5EA7) - App.tsx: use voiceLiveLightTheme/voiceLiveDarkTheme instead of webLightTheme/webDarkTheme - main.tsx: remove duplicate FluentProvider wrapper (App.tsx handles it) - Now --colorBrandBackground resolves to purple, not Fluent default blue - Start session button, orb, and all brand-colored elements are purple Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * feat(frontend): align UX with Foundry Portal reference implementation - CSS Grid layout matching agentPreview.module.css (items 1, 10, 11) - Background tokens using colorNeutralBackground3Selected (item 6) - Global CSS cleanup: removed conflicting non-Fluent rules (item 13) - Waves: dark/light theme support, fade-in animation (item 5) - VoiceOrb: 3-ring circleStack system (244/182/138px) (item 2) - SessionControls: 200px branded pill end button, brand mic (item 3) - TopBar: removed border, h1 ellipsis, disable New chat in session (item 4) - Idle state: proper spacing, start button sizing (item 7) - SessionEndedView: branded pill New chat button (item 8) - TranscriptOverlay: 420px max-width, 12px bubbles (item 12) - BuiltWithBadge: grid-positioned with container query responsive (item 10) Reference: ADO msdata/Vienna/azure-ai-foundry main branch Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix(frontend): match reference code exactly for controls, orb, and active session SessionControls: - CC: always transparent bg + neutralForeground2 (no brand-filled active state) - Mic: neutralBackground1 bg + brand border (not transparent) - Muted: neutralBackground3 + neutralForeground1 (not red/brand colors) - End: 20% width, min-200px, neutral bg + brand text + thin border VoiceOrb: - Active core: 160px with opacity 0.33 (was 120px) - Idle core: 120px solid (was wrongly 0.33 opacity) - stackTalking: reduces ring opacities (was scale animation) - captionsActiveHidden: hides circle when CC active ActiveSession: - Passes captionsActive to VoiceOrb for proper hiding Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix(frontend): match reference icons exactly TopBar menu: - Settings: Settings24Regular (was SettingsRegular) - Privacy: ShieldRegular (was ShieldCheckmarkRegular) - Terms/Privacy: Open16Regular external-link suffix (was ↗ text) SessionControls (action bar): - CC: ClosedCaption24Regular / ClosedCaptionOff24Regular (was custom SVG) - Mic: Mic24Regular / MicOff24Regular (was MicRegular/MicOffRegular) - End: Dismiss24Regular icon-only (was 200px text pill) - Icon color: colorNeutralForeground3 lighter at rest (was Fg2 darker) AgentPrimaryDetails: - Removed robot emoji icon (not in reference) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix(frontend): settings panel fields stacked vertically with full width fieldStyle now uses flex-direction:column so Label headers appear above their Textarea/Input controls, matching dropdown layout width. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix(frontend): session ended — add session ID box, remove New chat button and emoticons - Session ID displayed in bordered monospace box matching reference - Removed purple 'New chat' button (use header button instead) - Removed 👤/🤖 emoticons from transcript entries - Centered layout Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix(frontend): always show session ID box (with N/A fallback) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix(frontend): center agent details, expand CC transcript area, session ID box outline - Agent name + description: added alignItems:center to flex container - CC transcript area: removed justifyContent:center from container so transcript gets full available height, reduced gap to 12px - Session ID box: use colorNeutralStroke1 (darker) for visible outline Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix(frontend): color audit fixes — Fg2 icons, hover/active states, brand-blue token SessionControls: - Icon color: --colorNeutralForeground2 (#424242) matching reference line 472 - Hover: transitions to --colorNeutralForeground1 matching reference line 526 - Active: scale(0.92) matching reference line 530 ChatMessages: - User bubble bg: --colorBrandBackground (was undefined --brand-blue) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix(backend): forward service session ID from SESSION_CREATED event The VoiceLive service sends a SESSION_CREATED event with a ResponseSession object containing an 'id' field. We now capture this in _handle_event and include it as 'session_id' in the 'session_started' message sent to the frontend. Previously the frontend received no session_id, so the ended screen always showed 'N/A'. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix(backends): forward service session ID in all backends (JS, C#, Java) All backends now handle SESSION_CREATED to capture the service session ID, and include it as 'session_id' in the 'session_started' WebSocket message sent to the frontend. Pattern (same in all 4 backends): 1. SESSION_CREATED → store event.session.id 2. SESSION_UPDATED → fallback capture if not already set 3. session_started message → include session_id field Falls back to client_id if service doesn't provide one. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * feat(frontend): text mode improvements + audio playback toggle + agent mode guard Text mode: - Moved text chat + controls inside CSS Grid content area (same position as voice mode, no longer below page footer) - Added speaker toggle button (Speaker224Regular / SpeakerOff24Regular) with same circle design as mic button in voice mode - Audio playback defaults to OFF, user can toggle on - Dismiss X button for ending session (same as voice mode) Audio playback: - Added GainNode-based mute/unmute to useAudioPlayback hook - Exposed isPlaybackMuted + togglePlaybackMute through useVoiceSession Agent mode guard: - Agent mode toggle disabled in Settings when no agent is configured on the server (agentName and project both empty) - Matches main branch behavior Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix(frontend): text mode — wider controls, barge-in on send, skip audio when muted - Text input bar uses full chat width (removed justify-content:center) - sendTextMessage calls stopPlayback() for barge-in (interrupts audio) - playAudio skips decode+worklet entirely when muted (saves resources) - Uses isMutedRef for synchronous check in playAudio callback Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * docs: add URL query parameters, text mode, and session ID documentation - New 'Frontend URL Query Parameters' section documenting all 6 params (?mode, ?lock, ?agent, ?project, ?theme, ?greeting) - New 'Text Mode' section explaining Voice Live text modality (not Foundry Agent Service API), audio playback toggle, barge-in - Updated architecture diagram showing voice + text paths - Added send_text to WebSocket protocol table - Updated session_started to mention session_id - Added agent mode guard and audio playback UX guards Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix(frontend): voice mode audio no longer muted, text mode mute resets on new chat Root cause: useAudioPlayback had isPlaybackMuted=true by default and playAudio skipped ALL audio when muted — affecting both voice and text. Fixes: - Gain node starts at 1 (full volume) — voice mode always plays - Mute check moved to useVoiceSession: only skips audio in text mode - resetPlaybackMute() called on session reset (fixes stale icon) - Removed early-return from playAudio (was killing voice mode) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix(csharp): use _clientId instead of _config.ClientId for session ID fallback Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * feat(frontend): Foundry-aligned dark theme with custom background overrides theme.ts: - Dark theme uses Foundry Portal's HSL step system for backgrounds (bg1=12%, bg2=10%, bg3=8%, bg4=6%, bg5=3%, bg6=0%) - Brand foreground overrides for dark mode contrast (step 110/120/140) - Light theme uses step 70 for colorBrandBackground (darker, per Foundry) App.css: - Dark mode fallbacks now use Foundry-exact values: --bg-1: #1c1c1c, --bg-2: #1f1f1f, --bg-3: #1a1a1a --fg-1: #ffffff, --fg-2: #d6d6d6 --border: rgba(255,255,255,0.10) - Fixes settings panel white background in dark mode - Fixes all borders/overlays appearing as dark-on-dark Reference: ADO msdata/Vienna/azure-ai-foundry packages/core/contexts/theme/src/themes.ts Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * feat(frontend): switch to exact Foundry brand ramp + Aptos font Brand ramp: exact values from Foundry Portal themes.ts Base: #8251EE (was #7B5EA7) Step 70 (brandBg): #643FB2 (was #6B3AB3) All 16 steps now match Foundry exactly Font: Aptos, sans-serif (was Segoe UI Variable) Matches Foundry Portal's fontFamilyTokens Monospace: Aptos Mono (was system default) Reference: ADO msdata/Vienna/azure-ai-foundry packages/core/contexts/theme/src/themes.ts Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix(frontend): CC transcript uses full height — skip orb in CC mode When closed captions are active, the orb area (with hidden circle) was still consuming space via padding and flexShrink:0. Now the CC layout skips the orb entirely — transcript gets flex:1 full height, status text shown inline with minimal padding, controls at bottom. Normal voice mode layout unchanged. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix(frontend): text mode audio toggle works during active session Root cause: playAudio wrapper used isPlaybackMuted state in closure, which was stale after WebSocket onmessage handler was bound at session start. Toggling unmute created a new playAudio function but the WS handler kept the old reference with isPlaybackMuted=true. Fix: use isMutedRef.current (ref, not state) for the mute check. Refs are stable across renders and always reflect current value. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Add MCP config for Copilot coding agent Configures Azure MCP (docs, best practices) and Foundry Storybook MCP for the GitHub Copilot coding agent. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix MCP config: add required type field Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix MCP config: add required tools arrays Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Switch CodeQL to advanced setup with autobuild for C# (#23) Default setup used build-mode 'none' for C# which resulted in low analysis quality (call target resolution 83% vs 85% threshold). Switching to autobuild resolves NuGet dependencies properly. Other languages (JS/TS, Python, Java) remain on build-mode 'none'. Upgraded to security-extended query suite. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Add Dependabot config for automated dependency updates (#24) * Switch CodeQL to advanced setup with autobuild for C# Default setup used build-mode 'none' for C# which resulted in low analysis quality (call target resolution 83% vs 85% threshold). Switching to autobuild resolves NuGet dependencies properly. Other languages (JS/TS, Python, Java) remain on build-mode 'none'. Upgraded to security-extended query suite. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Add Dependabot config for all package directories Covers 28 directories across 4 ecosystems: - pip (5 Python dirs) - npm (10 JS/TS dirs including @azure/ai-voicelive) - nuget (6 C# dirs) - maven (2 Java dirs) Weekly schedule, voicelive-repo-maintainers as reviewers, language-specific labels for filtering. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Add send_text support to JS/C#/Java backends and fix useEffect deps Critical: Text mode chat (send_text WebSocket message) was only handled by the Python backend. JavaScript, C#, and Java backends silently ignored text messages, making text mode non-functional on 3 of 4 backends. Changes per backend: - JavaScript: Add sendText() using session.sendEvent with ConversationItemCreate + ResponseCreate - C#: Add SendTextAsync() using session.SendCommandAsync with raw JSON for conversation.item.create + response.create - Java: Add sendText() using typed UserMessageItem with session.addItem + session.startResponse Also fix React useEffect exhaustive-deps warning in App.tsx by adding all referenced values (agent, project, greetingDisabled, updateSettings) to the dependency array. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Add E2E test suite and fix Java SessionUpdate cast Add test_local_all_backends.py — comprehensive local E2E test suite that: - Starts each backend (Python/JS/C#/Java) sequentially on port 8000 - Tests WebSocket voice mode (send audio, verify response) - Tests WebSocket text mode (send_text, verify response) — validates the PR fix - Runs Playwright browser tests with video recording - Screenshots saved per backend per test phase Fix pre-existing Java compile error: SessionUpdate.getSession() doesn't exist on the base class. Cast to SessionUpdateSessionCreated/SessionUpdateSessionUpdated subclasses which have the getSession() method. Test results: 11/12 passed across all 4 backends. Only failure: Java voice WS test (pre-existing Spring WebSocket close frame issue). All 4 backends pass the critical send_text test. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix Java/Spring WebSocket close frame and message buffer issues Three fixes for the Java backend WebSocket handling: 1. WebSocketConfig: Add ServletServerContainerFactoryBean to disable idle timeout (0L) and increase message buffer sizes (512KB). Voice Live sessions stream audio continuously and need large buffers. 2. VoiceLiveWebSocketHandler.afterConnectionEstablished: Set per-session text/binary message size limits to 512KB for incoming messages. 3. VoiceLiveWebSocketHandler.handleStopSession: Send proper WebSocket close frame (CloseStatus.NORMAL) after the session_stopped response, ensuring clean close handshake with RFC 6455-compliant clients. 4. VoiceLiveWebSocketHandler.handleTransportError: Close WebSocket with SERVER_ERROR status on transport errors instead of leaving it dangling. 5. VoiceLiveWebSocketHandler.sendJson: Move isOpen() check inside the synchronized block and downgrade IOException to debug level (expected during teardown when VoiceLiveHandler reactor threads race with close). Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix Java interim text delimiter and move SDK calls off WebSocket thread Two fixes from Codex 5.3 review: 1. VoiceLiveHandler: Split interim_static_texts on newline instead of pipe character, matching Python/JS/C# backends and frontend behavior. 2. VoiceLiveWebSocketHandler: Dispatch sendAudio/sendText/interrupt to a dedicated executor thread pool instead of blocking the WebSocket I/O thread with .block() calls. Prevents stalling other clients under network latency. Executor is shut down in shutdownAll(). Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Add tests/results/ to .gitignore Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix CodeQL log injection findings (3 of 3) Sanitize user-controlled values before logging to prevent log injection: - Python app.py: Strip newlines/tabs from client_id (from URL path) - Python voice_handler.py: Strip newlines from service session_id - Java VoiceLiveWebSocketHandler: Strip control characters from clientId extracted from the WebSocket URL path Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Strengthen Java clientId sanitization for CodeQL Use allowlist pattern [a-zA-Z0-9\-_.] instead of blocklist stripping. Removes all characters except alphanumeric, dash, underscore, and dot to fully prevent log injection from user-controlled WebSocket URL paths. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds \dependabot.yml\ covering 28 directories across 4 ecosystems:
Config: Weekly schedule, \�oicelive-repo-maintainers\ as reviewers, per-language labels for filtering.