Изменения

• Auto Answer (Beta). Answers appear on their own the moment the other person finishes a question. Natively decides whether an answer is actually wanted, and stays silent when it isn't. Off by default; turn it on in Settings → General.
• A new answer engine. Answers now pick their source properly — your résumé, an uploaded document, the live conversation, or general knowledge — instead of blending them. Follow-up questions remember the thread even after the topic changes.
• Meeting notes, rebuilt. Summaries adapt to the kind of meeting you're in, get a real title instead of the first line of the transcript, and scale in length with the meeting. Plus editable speaker labels and a ready-to-send follow-up email draft.
• Natively browser extension for Chrome. Press ⌘/Ctrl+Y to pull the page you're looking at into your next answer — a job description, a coding problem, documentation. One click to enable it everywhere.
• Built-in Modes, including a new Call Center mode.
• Skills: upload and delete. Add your own skill as a file or folder with a preview before it installs, and remove ones you no longer want.
• NVIDIA NIM joins the provider list, with NVIDIA speech-to-text and nine streaming models to choose from.
• Role Insight (Pro). Paste a job description and see every requirement pulled out, checked against your profile, and scored for coverage.
• Windows: type into the meeting overlay without losing focus on your call. Works with Alt+Tab and clicking away, and steps aside for CJK/IME input.
• Redesigned meeting overlay. The panel, pill and resize control now move as one, answers stream in word by word, and scrolling back no longer fights you (with a jump-to-latest button).
• Parakeet CTC 0.6B added to the local speech models.
• LiteLLM gets a model picker and a configurable default.

• New default models. Gemini 3.1 Flash Lite and Gemini 3.7 Flash. The full Claude catalogue is now listed and correctly ordered, and the model picker shows real provider logos.
• Auto Answer and speech recognition are noticeably faster — answers are prepared ahead of time, and local speech no longer stalls when a meeting starts.
• Refreshed look throughout. Apple-style toggles, smoother section transitions, a new Soft Orchid accent, and a unified Plans & Billing page.
• Code blocks redesigned — clearer syntax colours, and the language label and copy button combined into one control.
• Calmer onboarding. Guided steps instead of a pile of notifications.
• Rarely-used settings tucked into an Advanced section.

• Startup. Fixed the app getting stuck on the logo, several memory leaks that could crash the launcher, and a crash that could stop the app opening ever again.
• Meetings were being lost. A database bug meant some meetings never saved. Fixed, and a meeting is never left unnamed.
• Microphone and audio. Fixed the mic being held after a failed start, capture breaking when you switch devices mid-meeting, and system audio not being picked up on some Windows setups.
• NVIDIA speech now stays connected through dropped streams, handles "Answer Now" without ending the session, and no longer goes silent when you remove the key.
• Answers. Your profile no longer leaks into unrelated technical questions, internal markers no longer show up on screen, and a slow provider can't overwrite an answer you already have.
• Language. Choosing English now actually keeps answers in English.
• Windows. Screenshots capture the right monitor, the cropper works across multiple displays, keyboard shortcuts no longer stop working in stealth mode, undetectable mode covers the tray and taskbar, and the overlay stays above fullscreen calls and screen shares.
• macOS. The Dock icon no longer reappears after stopping a meeting in undetectable mode.
• Chat. Typing on your desktop while a phone answer is streaming no longer swallows your message, and a failed phone answer no longer breaks desktop chat.
• API keys and licence. Keys survive a failed keychain save and a restored backup, corrupted settings no longer brick the app, and Pro stays active in more edge cases.
• Updates. A beta build can now update to the matching stable release.

• Electron 33 → 43, and TypeScript 7.
• Intel Macs now get a correctly built x64 DMG.
• Requires Node 22.6+ to build from source.
• Updated better-sqlite3, ws, react-syntax-highlighter, and @grpc/grpc-js; dropped sqlite3, react-query and other unused packages.

Note: Meeting notes now always use the new generator — the old toggle is gone. The new answer engine is on by default for everyone.

• Profile Intelligence Router (v2). Advanced domain classification (Coding, System Design, Behavioral, Negotiation) propagating constraints directly to LLM streaming paths.
• DeepSeek AI Support. Native integration of DeepSeek's advanced reasoning models via custom cURL OpenAI-compatible API providers.
• Two New Meeting UI Themes. Beautiful Liquid Glass and Modern Dark themes to completely redefine the real-time overlay visual experience.
• Answer-Type Constraints & Follow-Up Resolver. Context-aware follow-up resolution with strict output formatting layout constraints (short, detailed, bulleted, code-only).
• Eager Code UI Expansion. Growth-holds CSS elements to eagerly size overlays before React code-block mounting to prevent layout shifts.
• PI Latency Tracer (`PiLatencyTracer`). Telemetry to track reasoning, validation, and routing latencies to guarantee sub-500ms responsiveness.
• Evidence Validator & Live Deadlines. Cross-validates claims made in meetings and displays real-time countdowns for live assessment deadlines.
• Single-Click In-App Updates. Seamless update loops directly inside the desktop application.

• HTTP Keep-Alive Dispatcher. Integrated HTTP keep-alive connection pooling via undici agent to speed up API requests.
• WS DNS Cache. Resolved DNS lookup cache thrashing during key-rotation reconnect storms.
• STT Silence Watchdog. Extended silence watchdog reset triggers to Deepgram SpeechStarted events to eliminate false-positive probes during fast speech.
• Test Coverage. Added a 34-unit test suite (unit-fixes.test.mjs) to verify server logic.

• Audio Stack & TCC Permission Hardening. Hardened credentials management by eliminating racing set-provider IPCs and resolved macOS system audio process tapping/TCC permission gates to guarantee robust capture streams.
• ElevenLabs Failover Audio Gap. Resolved ElevenLabs open -> session_started audio gap on failover/reconnect (C1).
• Mic-Only Billing Bypass. Fixed mic-only billing bypass with active/recent system presence checks (C2).
• Stream-Abort Billing Leak. Fixed stream-abort billing leaks by moving billing triggers to the stream finally block (C3).
• Language Regex Prompt Injection. Patched language regex prompt injection security vulnerabilities on /v1/chat/completions (C4).
• Webhook Processing Retries. Implemented webhook processing retries with 3-attempt exponential backoff to prevent transient database-related data loss (C5).
• Fallback-Seconds Double Count. Fixed fallback-seconds double counting on STT reconnect-after-failover (C6).
• Admin Endpoint Sanitization. Sanitized admin endpoint provider-health key leak (H13).

• Added undici@^8.2.0 dependency to package.json.
• Exposed SILENCE_WATCHDOG_MS, SHADOW_PROBE_TIMEOUT_MS, and MAX_SHADOW_PROBES as test-overridable constants.
• Updated __tests__ suites with signature-tolerant assertions.

• Phone Link Integration. Connect your iOS or Android device to use it as a wireless remote microphone or a companion screen. Keep your meeting notes and AI suggestions on your phone while maintaining a clean desktop for screen sharing.
• TinyPrompts™ Engine. A ground-up redesign of our prompt architecture specifically for "Lower Tier" Small Language Models (SLMs) like Qwen 2.5:4B and Llama 3.2. These prompts are under 800 tokens, use a strict imperative voice, and eliminate XML overhead to ensure high-performance reasoning on local consumer-grade hardware.
• Codex CLI Integration. Native support for the Codex terminal-based automation system. Harness the power of gpt-5.3-codex for sandboxed code execution, local workspace tasks, and complex automation directly from the assistant.
• Auto-Calendar Sync. Natively now securely connects to Google Calendar and Outlook to automatically prepared context, participant details, and agendas before your calls begin.
• Speaker Identification (Diarization). Advanced real-time speaker tagging automatically identifies and labels individual speakers throughout your meeting transcripts.
• Smart Task Sync. Precision auto-extraction of action items with direct export support for Jira, Linear, and Asana boards.
• Advanced Stealth Mode. Hardened undetectable stealth mode with process name disguising, Activity Monitor evasion, and OS-level event blocking to ensure absolute invisibility.

• Deepgram SDK v3 & Nova-3. Complete migration to the latest Deepgram engine, resolving persistent 1006 connection loops and EPIPE errors.
• Intelligent Failure Detection. Introduced ElevenLabs "shadow probes" and a new STT silence watchdog that detects and recovers from silent provider failures in real-time.
• Deepgram VAD Integration. Leverages SpeechStarted events to eliminate false-positive silence detections during rapid conversations.

• macOS Audio Capture. Fixed CoreAudio Tap capture to correctly handle mono/stereo buffers, resolving the "pitched up" system audio bug.
• Streaming Resilience. Added AbortController propagation and 15s SSE heartbeats to prevent proxy timeouts during long AI reasoning tasks.
• Expanded Offline Mode. Now supports 100% offline transcription and note generation using specialized on-device SLMs.
• Language Detection Tuning. Improved batch detection confidence thresholds (0.65 min) to ignore background noise and silence.
• Stealth Evasion. Helper process renaming for improved stealth in Activity Monitor and strict management of forceUpdate timeouts during stealth transitions.
• UI & Navigation. Added new scroll keybinds for seamless, mouse-free interaction and optimized the horizontal layout for more stable code line rendering.

• STT Failover Logic. Fixed cascading failures where session teardown codes were incorrectly counted as health strikes.
• Trial System Persistence. Resolved a bug where the "Start Trial" card would reappear after expiry.
• Memory Optimization. Fixed exponential pre-buffer growth during STT key rotations that previously caused SSL/TLS parameter errors.
• Mic Activation. Prevented the microphone from eagerly activating in settings, which triggered the macOS orange mic indicator unnecessarily.
• Stealth Mode Stability. Fixed a regression where the macOS dock icon would mysteriously reappear after changing a disguise, and resolved macOS activate events accidentally breaking stealth mode.

• New `CodexCliService`. Supports multiple sandbox modes (read-only, workspace-write, danger-full-access).
• Optimized `tinyPrompts.ts`. Imperative, context-efficient prompts designed for 4B-8B parameter local models.
• Diarization Pipeline. Real-time speaker mapping and session-scoped identity persistence.

xattr -cr ~/Downloads/Natively-2.6.0-x64.dmg

`

2. Install the natively.dmg

3. Open Terminal and run: xattr -cr /Applications/Natively.app

• Modes Manager. Tailor Natively to your specific meeting context with 7 specialized personas: General, Sales, Recruiting, Team Meet, Looking for Work, Lecture, and Technical Interview. Each mode features custom system prompts and note templates.
• Custom Context. A new free-form textarea in Profile Intelligence—auto-saved and injected into all AI interactions—allowing you to provide persistent context like sales stats, tech stacks, or personal preferences.
• 10-Minute Free Trial. New users can now explore the full power of Natively, including premium STT and AI features, with a guided 10-minute trial and real-time quota tracking.
• Permissions Guided Setup. A new premium "Permissions Toaster" ensures Mic and Screen Recording permissions are correctly configured on first launch, reducing setup friction.
• Trial Promo Toaster. A subtle, beautifully designed delayed offer that introduces new users to the Pro features via a 10-minute hero countdown.

• STT Stability & Resilience. Completely rewrote the STT pipeline to use @deepgram/sdk v3 (Nova-3 model). Implemented a robust "shadow probe" system and smarter exponential backoff to eliminate 1006 connection drops and silent failures.
• Premium Upgrade Redesign. A full "Apple-tier" redesign of the upgrade modal, featuring clear plan tiers (Pro, Max, Ultra, Standard) with premium glass aesthetics and magnetic hover effects.
• Ad Campaign Rework. Optimized the ad engine to be less intrusive, with a 1h cooldown, sequential rotation, and plan-based targeting to avoid redundant offers for high-tier users.
• Diagnostic Logging. Added a new "Forward Log to File" capability (~/Documents/natively_debug.log) with an in-app toggle and progress bar for easier troubleshooting.

• Audio Privacy. Fixed an issue where the microphone could remain active while browsing the settings menu outside of a meeting.
• Screenshot Capture. Resolved a race condition that prevented screenshot listeners from attaching correctly during panel expansion.
• Trial Persistence. Fixed the "Start Trial" card incorrectly reappearing after a trial had already been claimed or expired.
• Branding Consistency. Completed the rebranding transition by replacing legacy "Cluely" references with "Natively" across all user-facing components.
• Pro Gating. Reinforced security gates on Modes and Profile Intelligence features to ensure correct license entitlement.

• Database Migration. Upgraded SQLite to v14 to support Modes, Reference Files, and Custom Context persistence.
• IPC Enhancements. Expanded the IPC layer with robust extraction for PDF, DOCX, and TXT reference files.
• Webhook Hardening. Refactored Dodo webhook handling with per-endpoint secrets and strict product-id validation to prevent accidental license provisioning.
• Deepgram & ElevenLabs Pools. Implemented multi-key pools with round-robin rotation and per-key health tracking for enterprise-grade reliability.

xattr -cr ~/Downloads/Natively-2.5.0-x64.dmg

`

2. Install the natively.dmg

3. Open Terminal and run: xattr -cr /Applications/Natively.app

• 10-Minute Free Trial System. Users can now fully audition Natively before purchasing, unlocking STT, AI processing, and web search quotas with seamless onboarding through a new countdown banner and promotional toaster UI.
• Guided Permissions Setup. Added a premium full-glass card experience on launch to smoothly handle Apple TCC permission requests for Microphone and Screen Recording.
• Premium Plan UI Redesign. Fully redesigned the plan selection and upgrade interfaces ("Apple/Tesla compact" look)—featuring glowing cards, hover effect physics, "Most Popular" and "Best Value" badging, and a clear BYOK path.
• Max & Ultra Upgrade Highlights. Pro users now receive beautiful gold/amber-themed dynamic upgrade comparison cards outlining the benefits of Max and Ultra tiers.
• Verbose Diagnostic Logging. Packaged apps now include advanced system logging capabilities—users can export natively_debug.log right to their documents via a new toast notification.

• Custom Provider Integration. Custom / cURL LLM provider configurations are now fully supported for meeting note generation, summarization, and follow-up suggestions—effectively sidestepping knowledge-mode interference.
• Data Privacy & Cleanup. Trial users who downgrade or expire now have their local profile data, resumes, and SQLite cache securely auto-wiped to maintain rigorous privacy policies.
• Advanced Audio Streaming. Upgraded the streaming session architecture to adapt to the new OpenAI Realtime Transcription API payload specifications.
• Ad & Campaign Targeting Engine. Re-engineered the in-app promotional system to intelligently swap visuals and logic strictly according to a user's exact active subscription level (planDetails), replacing random cycles.
• Global Rendering Shortcuts. Re-routed system-wide meeting screenshot calls (Cmd+H / Cmd+Shift+H) to leverage direct renderer interaction for rock-solid captures.

• Fixed a notorious background privacy issue where the system microphone would prematurely activate in the Settings page even without an active meeting.
• Resolved race-condition failures preventing taken screenshots from appearing natively inside the active meeting message bar layout.
• Fixed a bug where Profile Intelligence (such as resume analysis, company research, and negotiation hints) incorrectly blocked active Free Trial users.
• Corrected UI quirks during Free Trial initial loading—resolving a bug where trial-started "flashes" occurred upon mounting, and fixing stuck usage counters.
• Bypassed an invalid developer mode assertion preventing screenshot captures from resolving gracefully when app packaging scripts missed permissions.
• Resolved macOS cross-compilation errors (clang_rt.osx missing) by preferring the system Command Line Tools resource directory.

• Bound new ElectronAPI and IPC request handlers to securely coordinate: permission checks (permissions:check), audio tracking overrides, and local profile data annihilation (trial:wipe-profile-data).
• Handled telemetry logic mapping LLM __trial__ headers out of the main desktop API streams, converting them successfully into isolated x-trial-token usage patterns.
• Pushed major frontend dependency integrations for dynamic routing updates inside NativelyInterface.tsx and App.tsx state management trees.

xattr -cr ~/Downloads/Natively-2.4.0-x64.dmg

`

2. Install the natively.dmg

3. Open Terminal and run: xattr -cr /Applications/Natively.app

• Auto-Language Detection feature for seamless on-the-fly language identification and transcription routing
• Three new premium HTML automated email templates for transaction and marketing events
• Complete visual redesign of emails featuring a "Vivid Editorial" theme with soft glass gradients

• The interface now accurately remembers the overlay window position and handles multi-monitor environments
• Screenshots are now dynamically compressed via sharp before API transmission to prevent 413 Payload Too Large limits
• Hardcoded Railway URLs were replaced with the primary canonical domain for maximal uptime

• Fixed a race condition that could cause double-billing on Deepgram failover reconnects
• Eradicated critical server bugs ranging from ungraceful Railway restarts to Groq rate-limit memory leaks
• Resolved multiple security vulnerabilities including cross-user session theft and input validation flaws

• Shipped a comprehensive autonomous stress-testing suite simulating 30+ concurrent WebSocket connections
• Replaced Webhook logging arrays with optimized pre-allocated circular buffers to prevent memory exhaustion

xattr -cr ~/Downloads/Natively-2.0.2-x64.dmg

`

2. Install the natively.dmg

3. Open Terminal and run: xattr -cr /Applications/Natively.app

• Live Negotiation Coaching. Real-time salary negotiation advisor powered by a state machine that tracks offer phases, extracts amounts, and streams LLM-generated tactical guidance with a 5-second timeout and fallback copy
• Code Hint Mode (⌘6). Context-aware senior-reviewer hint streamed from screenshots; auto-extracts the problem statement on first press if no prior Solve was run; intelligently ages stale screenshot questions so multi-round interview sessions stay accurate
• Brainstorm Mode (⌘7). Streams 2–3 spoken, trade-off-aware approaches with brute-force → optimal pivot; integrates with the dynamic Recap/Brainstorm action button configurable per session from the Settings popup
• 4-Phase Interview Teleprompter. Replaced the old solution view with a structured Understand → Brainstorm → Implement → Verify teleprompter with spoken scripts, colour-coded phase cards, and complexity pills
• Auto-Expanding Code Panel. Overlay window smoothly expands to 1.3× width (780 px) when a code block is in view and contracts on scroll; driven by Framer Motion springs with zero re-renders per frame and a 2-call OS window sync strategy

• Dynamic Recap/Brainstorm button. Cmd+4 action is now a user-configurable toggle between Recap and Brainstorm modes persisted to settings and broadcast across all windows
• Natively API Promo Toaster redesign. Iridescent animated border (Aurora UI, 6 s loop), breathing glow, SVG grain texture, gradient headlines, and useReducedMotion gating for accessibility
• Full light mode system. Research-backed Professional SaaS palette (#F5F5F5 canvas / #FFFFFF cards / gray-900 text / blue-600 accent); all major components (Launcher, Settings, ModelSelector, JDAwareness, Premium, Profile, and Remote toasters) updated with depth-correct semantic tokens
• Company Intel Beta pill. Added Beta badge and disclaimer text to the dossier results section
• Profile Intelligence settings tab. Reordered navigation so Profile Intelligence appears immediately after Natively API
• Verbose debug logging toggle. Amber-accented terminal icon toggle in General settings; gates all audio pipeline diagnostic logs and STT segment sampling behind the flag
• Microphone device selection. Native Rust resolver honors user-selected microphone with exact-match → case-insensitive fallback → hard error listing available devices; "Default" sentinel check is now case-insensitive

• Focus steal eliminated. showInactive() used throughout toggle, screenshot-restore, cropper, and model-selector paths; setAlwaysOnTop only called when value changes; dock show/hide manages presence without calling focus()
• RAG global search. embedding_provider SQL filter was incorrectly excluding most meetings; fixed to ensure cross-session retrieval works correctly
• Silence timer bar freeze. Changed object-identity check m === msg to m.id === msg.id in the silence timer callback; spread-reconstructed arrays always invalidated the old reference before the 5-second timer fired
• Negotiation tracker cross-session bleed. JD replacement now correctly resets the negotiation tracker; deleteDocumentsByType only resets on RESUME delete, not JD delete
• Log rotation. 10 MB cap with .log.1 rollover prevents unbounded log growth on long-running sessions
• Concurrent audio test guard. _audioTestStarting boolean prevents two concurrent calls from both passing the async permission gate
• Fast mode persistence. Settings restoration moved to an unconditional startup path so the "Fast Response Mode" flag survives restarts
• Light/dark theme flash. Theme cached in localStorage and applied synchronously before ReactDOM.createRoot; body background changed from hardcoded #000000 to transparent to eliminate black flash in light mode
• Dark mode regressions. Launcher hero section, Settings rows, About creator pill, MeetingDetails tabs, FeatureSpotlight, and SettingsPopup toggle track all restored to original dark values after light-mode passes
• Code hint auto-screenshot. generate-code-hint and generate-brainstorm handlers now auto-take a fresh screenshot (restoreFocus=false) before reading the queue to prevent stale prior screenshots being sent to the AI

• CI smoke test. New build-smoke.yml workflow builds renderer + electron main process and asserts dist-electron/electron/main.js exists
• Native build hardening. build:native added to app:build pipeline; macOS defaults to current-arch only (NATIVELY_BUILD_ALL_MAC_ARCHES=1 for dual build); verifyArtifacts() throws on missing .node files (fail-fast)
• `electron/tsconfig.json` outDir. Corrected to ../dist-electron/electron; aligns tsc output with package.json "main" field
• Screenshot capture refactor. MacOS now routes all screenshots through desktopCapturer; session-based withScreenshotCaptureSession() pattern replaces ad-hoc hide/show callbacks; isCapturing guard moved to main.ts; cropper window now hides (not closes) on macOS to eliminate 1–2 s cold-start reload
• Keybind migration. chat:recap → chat:dynamicAction4 rename is handled by ID_MIGRATIONS in KeybindManager so user-customized binds are preserved across the update
• VectorSearchWorker. Integer validation (1–65 536) added for dim before table-name interpolation; invalid values post an error and break without executing SQL
• `embedding_provider` / `embedding_dimensions` written on summary embed. Prevents summary-only meetings from being silently excluded by provider-switch compatibility checks

xattr -cr ~/Downloads/Natively-2.2.0-x64.dmg

`

2. Install the natively.dmg

3. Open Terminal and run: xattr -cr /Applications/Natively.app

• Version bumped to 2.1.0 for the latest stable release.

• Improved compatibility with Dodo Payments API endpoints.

• Fixed a critical issue where valid Dodo Payments license keys failed to activate ("activation failed: status=201, err=unknown error") due to the system improperly handling HTTP 201 Created responses.
• Ensures all new premium license activations complete the checkout and activation flow smoothly.

• Rebuilt the Rust native module to correctly parse HTTP 2xx success statuses during activation checks.
• App binary repackaged and resigned for cross-platform distribution.

xattr -cr ~/Downloads/Natively-2.1.0-x64.dmg

`

2. Install the natively.dmg

3. Open Terminal and run: xattr -cr /Applications/Natively.app

• Dual-channel STT. System audio and microphone can now connect to the Natively API simultaneously using per-channel session keys (system / mic), eliminating concurrent_session_blocked errors
• Screen Recording permission flow. Proactive TCC dialog triggered via desktopCapturer.getSources() before the audio pipeline starts; clear error surfaced when permission is denied or restricted instead of silent black frames
• Dodo Payments integration. Webhook processing with HMAC-SHA256 verification, idempotent replay guard (24 h), subscription.on_hold / subscription.failed handling, and correct plan detection via product-ID env vars

• LLM resilience. 60 s / 3-retry wrapper for OpenAI and 90 s / 3-retry wrapper for Claude; Natively API falls back to Gemini on failure; streamChatWithGemini chain now tries Natively first
• GoogleSTT proactive restart. Pre-emptive stream restart at 4 min 30 s avoids the 5-minute hard-limit gap; improved buffer flush safety
• REST STT upload size. Audio resampled to 16 kHz mono before upload, cutting file size ~6× and keeping files under Groq/OpenAI 25 MB limit
• Deepgram reconnect cap. Max 10 reconnect attempts to prevent infinite retry loops during outages
• Windows z-order fix. setAlwaysOnTop re-asserted after showInactive() in switchToOverlay() and showOverlay() to prevent DWM silently demoting the overlay layer
• macOS compositor hide delay. Increased from 50 ms to 150 ms so the window is fully removed before screenshot capture, preventing black frames on slower machines
• STT key pre-population. Settings panel now correctly restores saved STT provider keys across all providers (Groq, OpenAI, Deepgram, ElevenLabs, Azure, IBM, Soniox)
• Custom provider timeouts. 30 s AbortController timeout added to executeCustomProvider() and streamWithCustom() to prevent indefinite hangs on unresponsive endpoints

• Fixed the audio transcription issue. Root cause was Deepgram receiving raw PCM without encoding=linear16 in the WebSocket URL, causing immediate upstream_closed with code 1000 and zero transcription output
• Fixed STT over-billing. Language/sample-rate reconnects produced short-lived sessions billed as 1 min each; added 30 s minimum threshold for sessions with no Deepgram speech, and changed Math.ceil → Math.round for fairer rounding; duplicate mic-channel billing that doubled every session is now removed
• Fixed native module crash. better-sqlite3 rebuilt with electron-rebuild for Electron's NMV 130, resolving NODE_MODULE_VERSION 141 vs 130 crashes that broke all DB operations (meetings, RAG, KnowledgeOrchestrator)
• Fixed asar stub audio silence. app.asar.unpacked path now checked first in nativeModuleLoader; added functional smoke-test (getInputDevices() must return an Array) to permanently catch this class of false-pass
• Fixed ad-hoc signing order. codesign --deep on the .app runs first, then .node binaries are re-signed with entitlements, preventing --deep from stripping the screen-capture entitlement
• Fixed keybind registration failure surfacing. When globalShortcut.register() returns false, a keybinds:registration-failed IPC event is now broadcast to the renderer so the UI can warn the user
• Fixed dark mode logo. force-black-icon CSS class now correctly renders the logo white/faded in dark mode; resolved specificity conflict where opacity was silently overridden
• Fixed theme flash on launch. Inline <script> in index.html reads localStorage and applies data-theme synchronously before the JS bundle loads, eliminating the light/dark flicker on repeat launches
• Fixed Natively STT fallback. Users with a stored sttProvider='natively' but no key no longer crash on meeting start; gracefully falls back to GoogleSTT
• Fixed `concurrent_session_blocked` permanent lock. Removed it from the fatal-error list; server closes the socket and scheduleReconnect() retries correctly after the old session is cleaned up
• Integrated several PRs. Napi-rs v3 migration for native audio entrypoints, Windows header layout fix, platform-aware shortcut symbols in About section, and various TypeScript type-safety improvements across electron.d.ts, preload.ts, and ipcHandlers.ts

• Migrated native audio entrypoints to napi-rs v3 ThreadsafeFunction API with err-first (err, arg) callback signatures
• nativelyapi/server.js comprehensive rewrite fixing 15 critical bugs (missing imports, ReferenceError constants, session key scoping, HMAC webhook verification, ElevenLabs message normalization, and more)
• scripts/build-native.js — dynamic Xcode clang version resolution fixes x86_64 cross-compilation linker error on Xcode 16+
• Added NSScreenCaptureUsageDescription and NSMicrophoneUsageDescription to mac.extendInfo for correct TCC permission dialogs
• nativelyapi/railway.toml deployment config added; production smoke-test script (test-prod.sh) included

xattr -cr ~/Downloads/Natively-2.0.9-x64.dmg

`

2. Install the natively.dmg

3. Open Terminal and run: xattr -cr /Applications/Natively.app

• Multimodal Groq Support. Integrated meta-llama/llama-4-scout-17b-16e-instruct into the ecosystem for high-speed screenshot analysis capability.
• Mouse Passthrough Mode. Merged backend Electron mouse event management with full state-sync between the global keybind manager and the React renderer interface.
• Updated Baseline Model Roster. Updated default architecture models to utilize the latest generation gpt-5.4-chat, gemini-3.1, and claude-sonnet-4-6.

• Instant Stealth Boot. Refactored the app initialization sequence to immediately read cached isUndetectable states on boot for an instant stealth startup.
• Windows Opacity Shield. Implemented an opacity shield on protected windows to eliminate a 1-frame screen flash when showing protected UI elements.
• Expanded Token Limits. Increased Groq max completion tokens to 8192 (the API maximum) to better support full code generation without hitting API constraints.
• Stronger Model Rotation Engine. Fortified the 3-tier fallback mechanisms and auto-upgrade logic across Gemini, Claude, GPT, and Groq models.

• Permanent Hide & Visibility Flaw. Repaired a critical IPC routing bug where hiding the session UI dynamically misrouted Cmd+B unhide commands to the background Launcher, eliminating the "invisible interface" trap that often forced application restarts.
• SQLite-Vec Corruptions. Fixed a critical silent data-corruption bug caused by tight strict dimensionality constraints (float[1536]), writing a multi-table database manager to ensure 100% of generated embeddings are retained and searchable.
• Groq Multimedia Drop. Resolved a "Front Door" routing bug where image attachments bypassed the Groq engine completely and triggered false LLM connection errors.
• Critical Race Conditions. Eliminated dangerous global state mutations during LLM fallback loops and patched SettingsManager early-initialization fatal crashes.
• Memory & Resource Leaks. Hardened native audio listener destruction to prevent zombie callbacks, and cleared dangling floating timeout IDs on opacity shields and disguise timers.
• Cross-Platform Compatibility. Rectified hardcoded natively.icns dependencies causing missing-icon errors on Windows/Linux environments.

• Native Module Loader Pipeline. Integrated a custom NAPI-RS absolute binary loader (nativeModuleLoader.ts) to entirely bypass POSIX-symlink require failures on Windows Git Bash.
• Safe Worker Instantiation. Relocated electron-specific module imports into execution-time lazy requires, neutralizing cross-boundary crashes.
• PR Code Integrations. Cleanly integrated complex changes from PR #64, PR #71, PR #113, PR #115.

xattr -cr ~/Downloads/Natively-2.0.8-x64.dmg

`

2. Install the natively.dmg

3. Open Terminal and run: xattr -cr /Applications/Natively.app

• Multimodal Groq Support. Integrated meta-llama/llama-4-scout-17b-16e-instruct into the ecosystem to enable native, blazing-fast screenshot analysis capability.
• Mouse Passthrough Mode. Added a system-level global hotkey to seamlessly toggle mouse pass-through, allowing direct click-through interactions with any background window beneath the assistant.

• Upgraded Model Roster. The baseline AI architecture now defaults to the leading-edge gpt-5.4-chat, gemini-3.1, and claude-sonnet-4-6.
• Model Rotation Engine. Fortified the 3-tier fallback mechanisms and auto-upgrade logic across all providers (Gemini, Claude, GPT, Groq) to guarantee absolute response generation reliability.
• Groq Token Expansion. Increased max completion tokens to 8192 (the API physical limit) to drastically improve full code generation output and prevent BadRequestError drops.
• Process Disguise Mapping. Enhanced stealth logic on Windows by contextually mapping macOS disguise names (e.g., automatically matching "Terminal" to "Command Prompt").
• SEO & Documentation. Deeply optimized the project README.md for search engines, injecting targeted keywords and detailed solution guides.

• Permanent Hide Trap & Data Loss. Repaired a critical IPC routing vulnerability where hiding the UI permanently trapped the app in the background. This previously frustrated users into Force-Quitting, which triggered data corruption and wiped out stored API keys.
• Vector Database Corruption. Fixed a silent indexing failure in sqlite-vec where using mixed embeddings (like local Ollama 768-dim alongside OpenAI 1536-dim) would permanently drop data. We've provisioned strict, dimension-specific tables (vec_chunks_768, 1536, 3072).
• Punctured Fallbacks. Corrected Gemini generateWithVisionFallback chains to properly execute tier downgrades instead of silently failing.
• Groq Multimedia Drops. Fixed a frontend routing defect that was preventing Groq from handling image attachments.
• Opacity Flash Bug. Implemented a Windows "opacity shield" to resolve a 1-frame visual screen flash when pulling up stealth/protected windows.
• Passthrough & Shortcut Dead Zones. Fixed bugs where the passthrough hotkey became temporarily unresponsive when the main app "stole" OS focus, and ensured the transparent overlay securely resets itself after ending a meeting.

• Native Audio Loader. Bypassed restrictive POSIX-symlink resolutions by mapping absolute paths directly to .node binaries, successfully repairing persistent start-up crashes—especially on Windows Git Bash.
• Memory & Listener Leaks. Eradicated massive runtime memory leaks by clearing unbounded array loops in the disguise engine, aggressively detaching Rust/C++ libuv event emitters on audio stops, and cleanly destroying opacity timers upon window focus shifts.
• Boot Race Conditions. Patched application boot order crashes involving userData lookups resolving before Electron was explicitly ready.
• Complete Thread-Safety. Eliminated dangerous global variable mutations in fallback chains by migrating to strictly localized modelIdOverride parameters, securing API loops.
• Settings State Persistence. Installed fail-safes into the SettingsManager parser so invalid or manually corrupted configurations default gracefully rather than immediately crashing the boot sequence.
• Dynamic OS Icon Binding. Ensured .icns, .ico, and .png icons dynamically bind to their respective Darwin, win32, and Linux target platform payloads.

xattr -cr ~/Downloads/Natively-2.0.7-x64.dmg

`

2. Install the natively.dmg

3. Open Terminal and run: xattr -cr /Applications/Natively.app

• Multimodal Groq Support. Integrated meta-llama/llama-4-scout-17b-16e-instruct into the ecosystem for screenshot analysis capability.
• Model Roster Update. Updated baseline architecture natively to default to gpt-5.4-chat, gemini-3.1, and claude-sonnet-4-6.
• Token Limits. Increased Groq max completion tokens to 8192 (the API maximum for context windows) to better support full code generation while preventing BadRequestError.
• Model Rotation Engine. Fortified the 3-tier fallback mechanisms and auto-upgrade logic for Gemini, Claude, GPT, and Groq models.
• OpenAI Streaming STT. Implemented a brand new low-latency WebSocket integration via the OpenAI Realtime API. Uses a 3-tier priority rotation (gpt-4o-transcribe → gpt-4o-mini-transcribe → whisper-1 REST) with server-side VAD, noise reduction, and uninterrupted audio buffering.

• SEO & Documentation. Optimized README.md for search engines with hidden targeted keywords.
• Code Quality. Performed a senior-level code review across modified files to address potential race conditions, edge cases, and empty references.
• STT Providers Architecture. Refactored Google, Deepgram, Soniox, and ElevenLabs streaming implementations. Specifically engineered the OpenAI module with custom ring-buffers, a 10s dark-drop timeout, a 5s zombie-session timeout, and 250ms audio chunk limiters to eradicate API rate-limits.
• Stealth Boot Refactor. Centralized platform disguise and dock icon management into AppState for consistent stealth behavior across reboots.

• Critical Race Condition. Eliminated dangerous this.geminiModel global state mutations during API fallback loops.
• Silent Fallback Failure. Rewired the generateWithVisionFallback chains for Gemini to correctly inject auto-discovered Tier models instead of defaulting back to generic UI settings.
• Groq Multimedia Drop. Repaired the "Front Door" routing bug in streamChat where image attachments bypassed the Groq engine completely and threw a "No LLM provider available" error.
• App Boot Race Condition. Wrapped SettingsManager.getInstance() constructor file-system access in app.isReady() checks to prevent early-import fatal crashes.
• Settings State Persistence. Added validation to SettingsManager's JSON parser so corrupted settings.json files default to {} safely.
• Opacity Shield Memory Leak. Saved references to the 60ms setTimeout Windows flash-shield timers in Windows Helpers and properly cleared them.
• Settings/ModelSelector Crash Risk. Appended .catch() blocks to the loadURL() directives in Windows Helpers to handle React dev-server drops gracefully.
• Disguise Timer Memory Leak. Rewrote main.ts's process title disguise implementation to immediately strip timer IDs when completed.
• Ollama Initialization Risk. Wrapped the OllamaBootstrap.bootstrap() floating promise in a tracked class property.
• Windows Icon Pathing. Rewrote the icon constructor option mapping to dynamically resolve natively.icns for darwin, icon.ico for win32, and icon.png for Linux.
• Cross-Platform Disguise. Verified mapping Terminal to Command Prompt on Windows and isolating CFBundleName safely to macOS environments.
• SQLite-Vec Per-Dimension Table Fix (v8 Migration). Fixed a critical silent data-corruption bug by provisioning three per-dimension table pairs (vec_chunks_768, 1536, 3072) and updating the VectorStore write path and native search payloads.
• Permanent Hide & State Clear Trap. Repaired a critical IPC routing flaw in WindowHelper.this.getMainWindow() where hiding the session UI dynamically misrouted all subsequent Cmd+B / "Toggle Visibility" commands to the background Launcher. This invisible interface trap caused users to repeatedly Force Quit the application via macOS dock. The forceful ungraceful exits during background syncing rounds led to truncated (wiped out) JSON files, erasing STT API keys and Disguise Settings. Atomic writes (implemented prior) prevent corruption during sudden exits, and this IPC fix completely solves the actual interface disappearance bug.

• PR Integration. Safely integrated changes from PR #64 ("Build stealth-mode enhancements") and PR #71, conducted code reviews, and ensured build compatibility without modifying git history.
• SettingsManager. Created SettingsManager to securely persist boot-critical settings (isUndetectable).
• Initialization. Refactored initializeApp sequence in main.ts to immediately read cached isUndetectable state on boot via SettingsManager for instant stealth. Migrated dock icon and process title management to a unified applyInitialDisguise lifecycle method.

xattr -cr ~/Downloads/Natively-2.0.6-x64.dmg

`

2. Install the natively.dmg

3. Open Terminal and run: xattr -cr /Applications/Natively.app

• Stealth Mode UI. The Process Disguise selector is now visually disabled and locked while Undetectable mode is active, preventing accidental state mismatches.
• State Synchronization. Greatly improved internal state synchronization across all application windows (Settings, Launcher, Overlay).

• Infinite Feedback Loops. Completely eliminated the bug where toggling Undetectable mode would sometimes cause the app to rapidly toggle itself on and off.
• Delayed Dock Reappearance. Fixed a regression where the macOS dock icon would mysteriously reappear several seconds after entering stealth mode if a disguise had recently been changed.
• Initial State Loading. Fixed an issue where the Settings UI would briefly show incorrect toggle states when first opened.
• macOS OS-level Events. Hardened the app against macOS activate events (like clicking the app in Finder) accidentally breaking stealth mode.

• Refactored IPC (Inter-Process Communication) listeners for SettingsPopup and SettingsOverlay to use a strict one-way (receive-only) data binding pattern.
• Added strict management and cancellation of forceUpdate timeouts during stealth mode transitions.
• Added explicit type safety for the new getters in electron.d.ts.

xattr -cr ~/Downloads/Natively-2.0.5-x64.dmg

`

2. Install the natively.dmg

3. Open Terminal and run: xattr -cr /Applications/Natively.app

• Two-Stage Silence Processing. Replaced basic RMS noise gating with a two-stage pipeline combining an adaptive RMS threshold and WebRTC Machine Learning VAD. Rejects typing, fan noise, and non-speech sounds before they bill STT APIs.
• Zero-Copy ABI Transfers. Transitioned the ThreadsafeFunction bridging to direct napi::Buffer (Uint8Array) allocations, completely eliminating V8 garbage collection pressure during continuous capture.
• Sliding-Window RAG. Implemented a 50-token semantic overlap in SemanticChunker.ts to prevent conversational context loss across chunk boundaries.

• Latency & Responsiveness Tuning. Stripped redundant TS debouncing, slashed MIN_BUFFER_BYTES, and reduced native hangover, achieving a ~300ms reduction in end-to-end transcription latency. short utterances ("Yes", "Stop") no longer sit trapped in the buffer.
• Removed floating-point division truncation for superior downsampling from 44.1kHz external microphones.

• Fixed a critical bug where the native Rust monitor returned a hardcoded 16000Hz while actually streaming 48kHz audio. Now syncs true hardware sample rates.
• Resolved the "Input missing" silent crash bug on microphone restarts by properly recreating the CPAL stream.
• Restored the 10s continuous speech backstop for REST APIs to prevent unbounded buffer growth.
• Added missing notifySpeechEnded() properties and cleaned up dangerous type casts.

• Audio processing transitioned entirely to strict ABI memory bridging (napi::Buffer)
• Re-architected native silence_suppression state machine around WebRTC VAD inputs.

xattr -cr ~/Downloads/Natively-2.0.4.dmg

`

2. Install the natively.dmg

3. Open Terminal and run: xattr -cr /Applications/Natively.app

• Dynamic AI Model Selection. Replaced static model lists with dynamic dropdowns. Your preferred models synced from providers (like OpenAI, Anthropic, Google) now automatically appear across the entire app.
• Multimodal Resilience. Added a "Smart Dynamic Fallback" using Groq Llama 4 Scout. If default vision models fail or get rate-limited during screen analysis, Natively instantly reroutes the image to ensure uninterrupted performance.
• Multiple Screenshot Support. The Natively Interface can now handle and process multiple attached screenshots simultaneously instead of just one.

• Improved Settings UX. API keys now auto-save after 5 seconds of inactivity, and selecting a preferred model immediately updates the app without requiring a reload.
• Better Embeddings. Migrated from Gemini Embedding to a completely new and more robust embedding architecture.

• Claude Fixes. Resolved max_tokens and context limits issues specific to Anthropic Claude interactions.

• DRY Refactoring. Centralized model configuration strings across the codebase to ensure easier future updates.

xattr -cr ~/Downloads/Natively-2.0.3-x64.dmg

`

2. Install the natively.dmg

3. Open Terminal and run: xattr -cr /Applications/Natively.app

• Fully functional system audio capture for Windows
• Introduced system for manual transcript finalization and interim/final bridging during recordings

• Migrated to app.getAppPath() for reliable cross-platform resource discovery
• Ensured sqlite-vec compatibility and fixed embedding queue management
• Upgraded @google/genai and optimized embedding dimensionality for lower latency

• Improved Soniox STT streaming reliability, manual flushing, and configuration persistence
• Resolved application entry point and module resolution issues in production builds
• Fixed transcript bridging for manual recording mode
• Corrected stealth activation and window focus inconsistencies

• Dependency updates for @google/genai
• Cleaned up native compiler warnings for Windows
• Fixed module resolution for internal Electron paths

• Live Meeting RAG Integration for instantly pulling facts from past meetings.
• Soniox speech provider for high-accuracy, low-latency transcriptions.
• Customizable AI response languages and speech languages.

• Radically redesigned Profile Engine UI with a mature Apple-like minimalist aesthetic and glassmorphism.
• Introduced Apple Premium Dark aesthetic for code blocks to preserve critical whitespace.
• Re-engineered Windows audio capture resolving WASAPI compatibility issues.
• Improved Markdown readability and stability across chat output via strict syntax highlighting.

• Resolved critical SQLite embedding indexing constraint errors.
• Fixed ad-dismissal cooldowns to properly track hidden states.
• Handled React dependency conflicts related to async-storage.
• Fixed Native module compilation and packaging for smoother delivery on Windows.

• Merged 3 major PRs streamlining repository contributions.
• Added correct Windows build icons.

• Multi-language support for STT and AI responses with language selection in settings
• Soniox STT provider added as a new speech-to-text option
• Google Custom Search integration for real-time research results
• Premium upgrade system with feature locking
• Enhanced Company Research Engine with JD context support
• Improved Knowledge Orchestrator and AOT (Ahead-of-Time) processing pipeline

• Launcher light mode theme refinements
• Better system audio device selection
• Improved Windows audio capture and WASAPI compatibility

• Fixed hardcoded English language for STT — Whisper now auto-detects languages
• Fixed OpenAI GPT 5.2 model compatibility (max_completion_tokens)
• Resolved merge conflicts and restored lost code changes across multiple modules

• Refactored knowledge pipeline and NativeAudio stubs
• Updated type definitions for stored credentials
• Dependency and build packaging fixes for production

• Replaced deprecated max_tokens parameter with max_completion_tokens required by GPT 5.x models.
• Increased max output tokens for OpenAI (GPT 5.2) and Claude (Sonnet 4.5) to 65,536.
• Increased max output tokens for Groq (Llama 3.3 70B) to 32,768.

• Improved response length capabilities across all text-generation AI models.
• Updated connection test model to use gpt-5.2-chat-latest instead of the deprecated gpt-3.5-turbo.

• Fixed 400 error when using OpenAI GPT 5.x models for text queries and toggle actions.

• Replaced max_tokens with max_completion_tokens in LLMHelper.ts and ipcHandlers.ts.

• API rate limiting to prevent 429 errors on free-tier plans (Gemini, Groq, OpenAI, Claude)
• Cross-platform screenshot support (macOS, Linux, Windows)
• Official website link added to the About section

• Smarter transcript memory management with epoch summarization instead of hard truncation — no more losing early meeting context
• API keys are now scrubbed from memory on app quit to minimize exposure window
• Credentials manager now overwrites key data before disposal for enhanced security
• Helper process renaming for improved stealth in Activity Monitor

• Fixed V8/Electron entitlements crash on Intel Macs by including entitlements.mac.plist during ad-hoc signing
• Fixed process disguise not applying correctly when undetectable mode is toggled on
• Fixed usage array capping with dedicated helper method to prevent unbounded growth

• Added RateLimiter service (token bucket algorithm with configurable burst and refill rates)
• Added PRIVACY.md and SECURITY.md policy documents
• Refactored ad-hoc signing script with helper renaming and proper entitlements flow
• Version bump to 1.1.7

• Fast Response Mode. Super-fast text processing powered by Groq Llama 3.
• Local RAG & Memory. Offline semantic search and retrieval across all your past meetings.
• Speech Providers. Added support for 7+ speech providers including Google, Groq, OpenAI, Deepgram, ElevenLabs, Azure, and IBM Watson.
• Custom Key Bindings. Full control over global shortcuts to fit your workflow.
• Stealth Mode 2.0. Enhanced disguise options (Terminal, Activity Monitor) and "undetectable" dock mode.
• Markdown Support. Improved rendering in usage section for better readability of AI responses.
• Image Processing. Faster image analysis powered by sharp optimizations.

• Performance. Optimized image handling and reduced latency for multimodal tasks.
• UX Refinement. Verified and fixed focus stealing issues for a smoother experience.

• Fixed UI bugs related to window focus.
• Resolved minor styling issues in the usage section.

• Integrated sharp for server-side image processing.
• Optimized SQLite vector retrieval for local RAG.
• Unified speech provider interface for 7+ integrations.

• Native Speech Provider Support. Added Deepgram, Groq, and OpenAI speech providers.
• Custom LLM Providers. Connect to any OpenAI-compatible API including OpenRouter and DeepSeek.
• Smart Local AI. Auto-detection of available Ollama models for local AI.
• Global Spotlight Search. Toggle chat overlay with Cmd+K (macOS) and Ctrl+K (Windows/Linux).
• Masquerading Mode. Appear as system processes like Terminal or Activity Monitor.
• Improved Stealth Mode. Enhanced activation and window focus transitions.

• Natural Responses. Updated system prompts for more concise and natural responses.
• Conversational Logic. Reduced robotic preambles and unnecessary explanations.
• Performance. Improved UI scaling and reduced speech-to-text latency.

• No critical fixes reported in this release.

• Internal logic refinements for improved conversational flow.
• Updater and background process stability improvements.

• OpenAI (GPT-5.2) & Claude (Sonnet 4.5) Integration. Fully supported alongside Gemini and Groq.
• Smart Dynamic Fallback. Automatically rotates to the next available provider if one fails (Gemini → OpenAI → Claude → Groq).
• Relentless Retry. Retries the entire provider chain up to 3 full rotations to guarantee a response.
• Enhanced Quality. System prompts now use provider-native API roles for better adherence and formatting.
• Improved Multimodal. More stable image understanding across all vision models.

• Zero-latency token streaming (no buffering).
• Optimized error handling.