Changelog

Released 18 Nov 2026

v2.0.0Release notes

Features · 96

  • Report workflow stage outcomes through a thread-scoped MCP tool, rejecting stale or unrelated reports and waiting for native completion before advancing.
  • Keep workflow-generated threads out of normal navigation and handle their approvals and questions directly in the workflow monitor.
  • Run every workflow stage in a persistent thread, preserving bot identity, model, provider instance, memory, and work-stage connectors.
  • Choose a work stage as each stage’s rework destination, retaining earlier accepted results while repeating the affected stages.
  • Build workflows with unlimited ordered stages, compact editors, bot avatars, shared results, and approval gates that return rejected work to stage 1 by default.
  • Inspect workspace AGENTS.md and CLAUDE.md as rendered Markdown in the right sidebar’s Instructions tab.
  • Discover existing Claude and Codex projects during onboarding with provider logos and import all or selected workspaces.
  • View, add, edit and delete Nora’s workspace memories from the right sidebar’s Memory tab.
  • Explain the personal workspace on its first visit after onboarding, including how it differs from folder and repository workspaces, and remember dismissal across restarts.
  • Choose where a new thread runs from the branch picker in its composer: the checked-out branch, any of the project's worktrees, or a new worktree created on the spot; threads in a worktree stay in the project's thread list with its role and bots.
  • Add MCP connectors directly in Nora from Settings → MCP Servers: enter a remote server URL, sign in through your browser with OAuth (or use an API key header), and every Claude, Codex and ACP thread gets its tools, with Nora storing credentials encrypted, refreshing sign-ins automatically and showing each connector's status in thread details.
  • Add, sign in to and manage MCP connectors straight from the chat composer's new Connectors button, which shows a few of the thread's connector logos and lists both Nora's connectors and the ones your Claude, Codex or ACP runtime already provides, without opening Settings.
  • Limit a connector to chosen workspaces or bots, so one bot can get tools others don't and a connector can stay inside a single workspace; pick them by bot avatar and workspace icon in the final Availability step of the Add connector dialog (now split into Server, Authentication and Availability steps), or from the connector's menu in Settings or the composer.
  • Type # in a thread prompt to search and select open workspace issues by number or title, inserting a GitHub or GitLab issue reference.
  • Add an Issues workspace tool with search, sorting, expandable descriptions and direct access to the full issue viewer for GitHub and GitLab.
  • Install the bundled Nora skill from Settings → Skills and hand tasks to agents by stable ID, with built-in tools to read, start, reopen and finish tasks while keeping completion status and board columns in sync.
  • Browse workspace MRs and pull requests in an edge-to-edge grouped full-screen table with search, sorting, expandable descriptions and changed-file previews, sidebar pinning, and access to the existing full-screen review viewer.
  • Change your work role from Settings → General using the original role picker, with your current choice selected, custom roles, and Save or Cancel controls; existing conversations keep their role.
  • Let Nora agents read and change any app setting through built-in get_settings and update_settings tools, including theme, appearance, integrations, AI, voice and thread defaults, with changes saved and reflected immediately in Settings.
  • Get a desktop notification when any thread or bot, including an asked bot or a group member working in the background, is waiting for your approval or answer, with the Dock badge counting what's waiting; clicking the notification opens that thread, a thread whose bot is waiting shows it on its sidebar row with a shortcut to the waiting thread, and while Nora is open, requests from other threads gather in a compact tray in the bottom-right corner, which collapses to a "waiting" pill, where you can allow, decline or answer them without switching threads.
  • Sync appearance, general preferences and thread defaults across signed-in devices with opt-in controls, saved-format migration, offline retries and per-setting merges; credentials and workspace data stay local.
  • Sign in, manage your profile and sign out from the sidebar avatar using compact Clerk dialogs styled to match Nora’s controls and theme, with encrypted session storage and optional local-only use.
  • Show a compact user avatar at the bottom of the Threads/Bots strip, using your name's initial with a user icon fallback and a larger click target.
  • Create bots with a name, role, instructions, model and an animated line-style avatar (a coloured outline whose eyes show what the bot is doing, matching withnora.run), everywhere or per workspace (private, or shared with collaborators as .nora/bots files), and message them from the Bots sidebar; the threads bots work in keep the instructions they started with and show the bot's face in the sidebar. Switch the sidebar between Threads and Bots from a vertical strip beside it to browse workspace and global bots and see the threads each one works in. A bot's work thread switched to another model offers a one-click way back to the bot's model.
  • Ask Nora, now listed first in the Bots sidebar as your built-in app assistant, to create bots for you from any thread with Claude, Codex or ACP agents that support HTTP MCP servers; Nora's built-in server and its tools are listed first in MCP Servers settings with whether it is running.
  • Nora threads can now update bots, list workspaces and each provider's models, list and read other threads, and list, create and move tasks on the workspace task board, all through Nora's built-in tools.
  • Ask Nora to create, edit and run workspace workflows with built-in tools, then check, pause, resume or cancel their runs; workflows it saves appear in the Workflows tool straight away.
  • Let Nora and bots ask other bots for help with ask_bot (including when you @mention a bot to Nora, with bots suggested as you type @): the asked bot answers in its own read-only child thread with its own model, each ask shows as a card with the reply and a link to that thread, and chains of bots asking bots are limited and cannot loop.
  • Give Nora and bots memory that carries across threads: they save durable notes with remember, correct or merge them and forget stale ones, find them with full-text recall search, their newest notes are added to new turns as context (with a pointer to recall when older ones don't fit), bots can list and read their own earlier threads to pick up past work, and you can see, add, edit and delete notes in each bot's settings or info panel, including Nora's.
  • When Nora or a bot saves a note with remember, the thread shows an animated "Remembered" line with a brain icon and the note, pulsing while it saves and drawing across the thread once it is kept.
  • Nora and bots now learn from their threads automatically: when a thread goes quiet, is settled or reaches 10 turns, a background review saves lasting preferences and facts, corrects outdated notes and shows them as "Saved automatically", never changing notes you wrote. Choose its own source and model (Claude or Codex runtime, or Cloud API) under Memory in Settings → Bots, turn it off there, and see what each review changed and how many tokens it used over the last 7 days.
  • Start group chats with 2–6 bots from the Bots sidebar: @mentioned members (suggested as you type @ in the composer) (or the ones Nora picks) reply one at a time in the group from their own threads, can hand off to each other by @mention, and only the member you choose can change files.
  • Chat with each bot in one persistent direct message, and with groups, in a messages-only chat: replies arrive as chat bubbles with a typing line while the bot works, approvals and questions appear as cards in the chat, each new task runs in its own work thread (in whichever workspace you choose under "Working in", or the one the message names), shown as a tab beside Chat at the top of the conversation with whether it is working or waiting on you, where you keep chatting with the bot inside that thread only; branch off into a side chat with + (a blank branch, or pick a group member) or Branch on any reply (carrying the chat up to that message), each branch named by the simple AI tasks model (from the message it branched from, or a blank branch's first message), with a panel beside each branch showing the message it was branched from, who it is with, its workspace, model and status, so you can run several conversations with one bot without mixing their context, close a branch or task tab with × and confirm to settle it (reopen it from the sidebar), and the main chat never routes its messages into your branches; every reply's View work link opens its work thread in full with its details panel.
  • Give group chats an optional lead who answers messages Nora can't match to a member, ask every member at once with @everyone, and let members read earlier group messages with a new chat_history tool instead of losing them from a longer, 40,000-character catch-up.
  • Show or hide bots' work threads from Thread view options, the eye button beside Groups in the Bots sidebar, or the one in a chat's header; hidden threads still appear when they need your approval or input.
  • In development builds, open a Next request context panel from a bot DM's header to see each model request your next message would make (the router and the work thread, switchable between open tasks and new ones), part by part with the exact text, estimated tokens and the thread's last measured input, cached and output tokens.
  • Rename groups and threads, or settle and reopen groups, from a right-click menu in the Threads list and the Bots sidebar's Groups section.
  • Open a bot's info panel from the info button or right-click menu on its row in the Bots sidebar (which also offers Message, Show work threads and Edit bot), its card on Bots home, or by clicking its face or name in a chat header, chat message, ask card or work thread, to see its model, availability, instructions, notes, recent work and groups, and message it or edit it from there.
  • Delete a bot from its right-click menu or info panel in the Bots sidebar, after a confirmation that warns when a shared bot's file will be removed from the workspace.
  • Add a Bots home, opened from the Bots strip, with cards for Nora and each bot (start or continue threads, edit), your group chats, an introduction to bots when you have none, and twelve templates such as Code reviewer, Meeting summariser and Planner that open the bot editor prefilled, with the ones suited to your workspace's role suggested first.
  • Budget supported Claude text tool results before model delivery with full local output, diagnostic excerpts and targeted reads, and use Codex's native output ceiling without increasing stricter configured limits; configure this in Threads settings.
  • Add a Tool output budget dev simulator scenario with a large synthetic log, a retained error excerpt and a targeted read of its saved output, without using model tokens.
  • Restore detected npm, pnpm, and Yarn commands on workspace home in a compact searchable picker with a Run button that opens a workspace-root terminal.
  • Add a home button beside the workspace picker to return to workspace home while keeping thread tabs open.
  • Change providers or model settings in an idle chat on the next send, preserving visible history and continuing in a fresh session with a bounded handoff and on-demand history reference.
  • Add a Switch provider dev simulator scenario with local Codex and Claude sessions, saved handoffs and a simulated on-demand history read, using zero model tokens.
  • Attach native chats beside Attach files as dated references with paginated history and bounded text chunks, loading their contents only when the agent reads them.
  • Open Threads settings directly from a cog in the chat composer's top-right corner.
  • Set a default subagent model for new threads in Threads settings, with per-thread overrides and fallback when the provider or model is unavailable.
  • Choose whether follow-up messages queue or steer a supported active turn, preserving delivery confirmation and retry safety.
  • Turn automatic thread title generation on or off in Threads settings.
  • Automatically move completed, inactive threads to Settled after a configurable period, preserving history and excluding pending work.
  • Configure Claude's context compaction threshold in Threads settings while retaining provider defaults when unset.
  • Add Threads settings with compaction suggestions for large native chats after estimated cache expiry and opt-in Claude compaction before expiry that skips active work and pending approvals.
  • Show native chat cache reuse and reported cache writes in context details, and retain cache reads and writes in provider usage accounting.
  • Bound attached archived history to a small selection of intact entries, preserving the full selected context in local files the agent can read on demand.
  • Attach images and files to chat by pasting, dropping, or choosing files, with image thumbnails and removable attachments preserved in drafts and delivery retries.
  • Settle idle threads to hide them without deleting history, then browse Settled threads and reopen them when needed.
  • Discover local Codex and Claude CLI and desktop coding conversations, resume them with saved history and native context.
  • Show workspaces together with colorful two-letter avatars and user-selectable colors, plus saved Workspace, Activity, or flat thread grouping in the sidebar.
  • Add Claude's Auto permission mode and distinguish Manual, Accept edits, and Plan with descriptions of how approvals work.
  • Show Claude's suggested follow-up as ghost text in an empty chat composer, with Tab to accept without sending and a simulator scenario for trying it.
  • Add a Stacked approvals simulator scenario that raises three requests together for testing approval, denial, and stopping the task.
  • Add a Background requests simulator scenario that raises approvals and questions while you are in another conversation, for trying the in-app request toasts without model tokens.
  • Run installed ACP coding agents such as Cursor, OpenCode, Google Antigravity, Grok, Copilot, Goose and Kimi as native thread providers, with their own models and modes, approvals, interrupts, sign-in status and provider icons, connecting agents installed while Nora runs without a restart.
  • Show whether Claude and Codex are signed in, with the plan and account, on each provider in Providers settings, and sign in from there through each runtime's browser sign-in.
  • Discover MCP servers and tools from Codex and Claude runtimes (ACP agents that do not report MCP status are left out instead of listing a notice for each) in a dedicated MCP Servers settings table showing each server's own logo and provider icon, global or workspace scope, connection status and tools, with refresh, browser sign-in for Codex and Claude servers that need it (including claude.ai connectors) without leaving Nora, and an explanation when Claude account connectors cannot load.
  • Generate commit messages and new thread names using signed-in Codex or Claude runtimes, API keys, or local models.
  • Show a PR icon with its number and state in the tooltip for sidebar threads linked to a branch pull or merge request, alongside activity and attention indicators.
  • Show how full the thread's context window is as a donut beside Send for Claude and Codex, including existing and resumed threads read from saved history, with token details and a Compact context button in a hover popover.
  • Show compact stacked toasts in the bottom-left on launch, each with the provider's icon, for every installed provider CLI (Claude Code, Codex, and other npm-distributed agents) that is behind its latest release, with Update and Dismiss buttons; dismissed updates stay available from an icon at the bottom of the sidebar until the next launch.
  • Show a Nora icon at the bottom of the sidebar while a Nora update is downloading, ready to install (click to restart and install), or available as a Linux release.
  • Show the selected provider's weekly and 5-hour usage as a double donut beside the composer's model picker, with limits, reset times, plan details and refresh in a hover popover.
  • Show a connector sign-in card in chat when the model asks you to sign in to an MCP connector, with sign-in, retry and live connected status shared with thread details, plus a Connector sign-in simulator scenario.
  • List a thread's running and finished Claude and Codex subagents in thread details with the model and effort each ran on, each expandable to its task, live progress, result and its own tool calls, plus a Subagents simulator scenario.
  • Show the skills a thread loaded or read in thread details, marked Loaded (reported by Claude) or Read (from the skill's SKILL.md), with the agent and calls that used each and the rest of the runtime's skills listed as available, plus a Skills simulator scenario.
  • Choose a default model for new threads in Threads settings, so new threads and drafts without their own model choice start with that provider and model, falling back to the first available one if it is removed.
  • Choose a default subagent model for new Claude threads in the model picker, shown beside the thread's model and reasoning.
  • Add a provider usage-limit simulator scenario to preview an exhausted provider account and example reset time in the chat and sidebar.
  • Show the thread checkout's branch diff against its default branch (+/− lines and files, including uncommitted work) in floating thread details, with View diff to open it in the built-in diff viewer and a Create PR or Create MR button for the connected GitHub or GitLab integration that pushes from the thread's own worktree.
  • Steer queued messages into an active turn when the configured runtime supports it, with delivery confirmation and protection against duplicate submission.
  • Add a streamed Markdown gallery to the chat simulator for testing headings, emphasis, lists, tables, quotes, links, task lists, and code blocks.
  • Enable structured chat questions and present them as selectable choice cards with single- and multi-select support, keyboard controls, and complete-answer submission.
  • Add an isolated developer chat simulator for exercising streaming, tools, questions, approvals, queues and recovery without LLM tokens or real tool execution.
  • Add a floating thread-details panel with reviewed file and line changes, branch PR/MR status, and archive controls in place of each thread's header row.
  • Show the MCP connectors enabled by a thread's provider in thread details, with each connector's logo, name and tool count or status, the first few inline and the rest in a Show more popover, in a taller details panel with a visible scrollbar, and sign in to connectors that need it from their row.
  • Replace separate managed-agent and AI-chat entry points with native threads in the existing workspace shell, while retaining manual terminals and workspace tools.
  • Prepare unified thread navigation, tabs and a direct chat composer that restores independent drafts and safely retries unconfirmed instructions.
  • Prepare a provider-and-model popover for threads, with provider icons, native model catalogs and selection that preserves drafts and permissions.
  • Prepare role-based onboarding, app defaults and workspace overrides, a no-repository Personal workspace, and guidance saved per conversation for Nora’s existing interface.
  • Show a live status line under a running turn so gaps between tool calls are explained: Thinking… while the model reasons (even when its thinking is hidden), Starting… and Stopping…, or a light filler phrase like Mulling it over… while it is between steps.
  • Give workflows an optional description, edited alongside the name, shown in the Workflows list and shared with every stage when a run starts.
  • Choose each workflow stage’s provider instance and model from labelled pickers, with provider logos or bot avatars in workflow summaries and execution through the selected thread runtime.
  • Give bots routines: several scheduled instructions per bot (hourly, daily, weekdays, weekly, monthly, every N minutes/hours/days, once, or a custom cron expression, on a chosen timezone), set up in a wide step-by-step editor (template, task, schedule, delivery) starting from a blank routine or one of seven templates such as a morning briefing or weekly summary, each run working in the routine's chosen workspace and posted to the bot's chat or kept in the routine's own thread, with pause, resume, Run now, a choice to skip or run once the runs missed while Nora was closed, and recent run history in the bot's info panel.
  • See every bot's routines on one calendar from Routines in the Bots sidebar, with month, week and day views showing past runs by outcome and upcoming runs from each schedule, and a details popover to open a run's thread, edit the routine, run it now or pause it.
  • Show interactive widgets from MCP tools (MCP Apps) inline in Claude and Codex threads, themed to match Nora and sandboxed on their own origin; in Codex threads widgets can also call their server's tools, and messages they suggest go into the composer for you to send.

Fixes · 63

  • Turn on settings sync again after updating Nora: saved sync state from earlier versions now loads when synced settings are added or removed, and enabling sync says whether you need to sign in or wait for settings to load.
  • Package Nora Desktop without the bundled mobile app and remote Worker projects, fixing macOS packaging failures and keeping the app download small.
  • Keep typing responsive during background updates by reusing snapshot projections and rendering conversation results at lower priority.
  • Stop idle agent-attention polling from repeatedly rendering the entire app shell after startup.
  • Keep composer typing local and batch draft saves, flushing before send, while avoiding redundant tab, model, and transcript updates.
  • Keep workflow cancellation final during checkout and thread preparation, and serialize run controls with execution updates.
  • Preserve live setup status when workflows are read or refreshed instead of incorrectly reporting an interrupted run.
  • Accept workflow results only from successfully completed threads; failed, interrupted, and timed-out stages cannot advance the workflow.
  • Bound each workflow turn by the remaining total runtime and reject results returned after the deadline.
  • Enforce read-only workflow reviews with Claude inspection tools or Codex sandboxing, rejecting providers without a supported review mode.
  • Notify users when a Nora app update is available and require choosing Download update before staging it for installation on restart. Preview the release notes in an expandable What’s new section before downloading.
  • Show Nora’s Everywhere memories alongside workspace-specific notes in the Memory sidebar, including Personal.
  • Discard draft threads from their right-click menu (asking first when they hold unsent text or attachments), and closing an empty draft's tab now removes it instead of leaving it in the thread list.
  • Show dialog field labels and typed values at full contrast, so the Add connector form and other dialogs no longer look dimmed.
  • Fix menus, popovers and dropdowns opened inside Settings and other dialogs ignoring clicks, which then fell through to the page behind them.
  • Keep long threads cheap when Nora's or a bot's notes change: a thread keeps the notes it started with in its instructions and receives later changes with your next message, so Claude and Codex no longer re-read the whole conversation uncached, and ACP agents now see note changes mid-thread.
  • Refresh task lists and boards immediately when an agent creates, moves, starts or finishes a task.
  • Make your messages in bot and group chats readable in light and dark themes with a soft accent-tinted bubble instead of a solid fill that hid the text.
  • Hide the edited-files card at the end of a chat turn when no files changed.
  • Submit task and planning prompts when starting their new threads instead of leaving them in the composer.
  • Open a dedicated branch-name dialog from New worktree and create a managed checkout without opening a thread, agent or terminal.
  • Keep sidebar footer icons on one row when Nora or provider updates are available, including at narrow sidebar widths.
  • Return the MR detail viewer’s Back action to the workspace MRs table, preserving its filters, expanded row and open review tabs.
  • Render sanitized HTML in pull and merge request descriptions and comments, including linked deployment badges.
  • Make workspace Home and Bots home replace an open merge request, file, diff or split view while preserving open tabs and thread drafts.
  • Open task planning and task work through native threads without obsolete CLI selectors or installed-CLI requirements, and keep planning errors in the dialog.
  • Count Codex usage per turn instead of repeating lifetime thread totals, read native cache-write counts correctly, and leave unmeasured resumed-turn usage unknown.
  • Keep thread settings provider and model dropdowns hidden while their options load, with explicit loading, empty, and unavailable states.
  • Keep personalization choices such as roles and the default thread model when other settings are changed, instead of reverting them to an older copy.
  • Load a thread's skills and MCP connectors in thread details on the first try instead of only after a manual refresh, by discovering them only for threads you open rather than every restored background tab, and sharing one Claude Code start across a thread's discovery reads.
  • Combine back-to-back reasoning in a thread into one collapsible Reasoning block instead of a separate block for each step.
  • Stop showing an empty Reasoning toggle in threads when a model reports reasoning with no visible text.
  • Open threads started outside Nora on the first click instead of only loading them until a second click opened a new tab.
  • Fix the app crashing when opening some threads by updating Electron to 42.11.18, which corrects a string-length bug in Electron 42.3.1's main process.
  • Label Codex's five-hour usage limit as 5-hour instead of Hourly in usage panels.
  • Fix Codex external-thread discovery failing when saved sessions have no recorded originator.
  • Prevent recurring startup checkout warnings after macOS remounts by using stable directory identities and refreshing workspace selections while preserving conversation checkout checks.
  • Offer workspace grouping only for All workspaces, hiding it in This workspace and restoring the saved grouping when switching back.
  • Skip window notifications after the renderer frame is disposed, preventing shutdown progress and late output from logging send errors.
  • Prevent the standalone chat simulator from going blank on the first message when thread details open.
  • Present approvals in a compact floating dialog stack with single-line action buttons, a clear primary action, scope tooltips, and dismiss/reopen controls, advancing one at a time for the active thread.
  • Open file links in chat in the built-in file viewer instead of navigating away from the app.
  • Keep the new-thread greeting and name together by using the available width and preventing the name from wrapping onto a line alone.
  • Hide floating chat details and reclaim transcript space while the changes sidebar is open, restoring the previous detail view when it closes.
  • Restore clicks and keyboard interaction in the GitHub device-code login dialog when opened over Settings.
  • Match active workspace and thread tabs to the content surface in dark mode, keeping a clear selection indicator and consistent hover shading.
  • Preserve thread-specific change counts when unrelated workspace files exceed capture limits by fingerprinting large files instead of discarding the entire turn snapshot.
  • Keep simulator scenarios and threads independently scrollable, and prevent streaming replies from pulling the transcript back down while reading earlier content.
  • Restore chat Markdown styling for headings, lists, quotes, links, tables, and code blocks, including plain fences and table alignment.
  • Show green additions and red deletions for file activity, per-file diffs and turn reviews, and include sample diffs in the developer simulator.
  • Allow developer simulator scenarios to run while other simulated chats await answers, approvals or recovery, without checkout write conflicts.
  • Keep provider failure messages visible in the active chat after a task ends or loses its connection.
  • Keep native execution available after closing and reopening the macOS window, and distinguish connection failures from setup save errors.
  • Keep startup working when saved workspaces change branch or become unavailable, preserving checkout and conversation identity checks.
  • Restore development builds after the native-thread migration by correcting browser review targets and removing stale renderer bindings.
  • Wait for startup and IPC registration before opening the window on macOS activation, preventing missing personalization handlers on first boot.
  • Use Nora's icon in the Linux dock by matching development window identity to its launcher and including the app icon in Debian packages.
  • Blend the chat input into its composer surface without a separate background or inner border, including hover and focus states.
  • Keep draggable window chrome and minimize, maximize, and close controls available during role onboarding, startup, and connection errors.
  • Prevent hidden workspace tab panes from receiving keyboard focus while preserving their mounted content.
  • Keep Codex model discovery compatible with runtime-advertised reasoning options and validate selections against the chosen model.
  • Report actionable Codex authentication and runtime-readiness failures before submitting a turn.
  • Choose which checkout the Changes panel shows from a picker in its header — the checked-out branch or any worktree, with the threads working in each — so it no longer jumps to a workflow run's worktree or follows whichever agent or terminal is focused.

Improvements · 91

  • Nora can now read and change the "Keep running in background" and "Launch at login" settings when you ask it to in a thread.
  • Sign in to Nora Desktop, the web app and remote access with Nora's production accounts on withnora.run (testing accounts from the preview do not carry over), and keep remote computer addresses API-only so they can never show pages or set cookies on Nora's domain.
  • Redesign the mobile and web app to feel like a native iPhone chat app: a ChatGPT-style home with the full composer and project menu, a drop-down computer switcher in the top bar, a slide-out sidebar for chats, projects, approvals and activity, Stop in the composer, iOS-style lists, sheets and alerts, and neutral system light and dark colours.
  • Reorganize the workspace actions menu into labelled Create, Terminal, Workspace and Close and remove sections with consistent "workspace" and "New thread" wording, scripts and terminal presets in pop-out submenus, new New worktree, pin, avatar colour, Reveal in folder and Copy path actions, and a confirmation before removing a workspace.
  • Give every dropdown, context menu, select list and suggestion list one shared, roomier style with larger text, quieter icons and cleaner section dividers, and turn long recent-workspace and worktree script lists into pop-out submenus.
  • Simplify sidebar usage with visible limits, slim progress bars, and account actions tucked into provider options.
  • Remove the legacy Context sidebar tab; archived conversation history and imported context files remain available.
  • Redesign the right sidebar to match the new app look: underlined section tabs that no longer truncate, a composer-style commit box with compact pull, push and PR actions, quieter commit and file lists that lead with the file name, and lighter Forge and Context views.
  • Use a project's own logo, app icon or favicon from its folder as its workspace avatar in the switcher and thread groups, falling back to coloured initials when none is found.
  • Show your account photo as the Personal workspace's avatar when signed in, or an abstract gradient artwork otherwise, instead of coloured initials, and repair Personal workspaces that older versions saved as plain folders.
  • Show the bot's avatar (or a group's stacked avatars) on bot chat tabs in the workspace tab strip instead of the generic chat icon.
  • Restyle the Workflows tool to match the Bots home: a centered list of saved workflows and runs with status, simpler create and run wizards without tinted panels, and a cleaner run monitor with progress, activity and output side by side.
  • Shimmer the live activity rows in a thread while the model is running tools or subagents, so in-progress work stands out from finished steps.
  • Choose which Claude or Codex model simple AI tasks use in Settings → AI instead of always the runtime's default, and see provider logos next to every provider and model in AI and memory settings.
  • Widen sidebar bot rows with smaller side gutters, larger avatars, and more vertical padding.
  • Render known @ bot mentions as avatar-and-name pills in the composer, with atomic editing and unchanged mention text when copied or sent.
  • Render selected issue references as inline GitHub or GitLab pills in the prompt composer, preserving their full links when editing, copying or sending.
  • Redesign issue details around native fix threads and compact GitHub or GitLab issue reference pills instead of pasting issue contents into prompts.
  • Start an individual task directly from its list row or board card without selecting its checkbox.
  • Offer an optional Nora skill install during onboarding, recognising existing installations and allowing users to retry or skip until later.
  • Create tasks in a dialog with title, goal, context and completion criteria, saving the Markdown file only when submitted.
  • Bring diff and pull-request reviews, document and task panels, deployment views, Settings and utility dialogs in line with the current app styling, using shared corners, neutral surfaces, readable labels and accessible comment actions.
  • Refresh browser review and annotation controls with compact chrome, neutral theme surfaces, readable comments and keyboard-accessible actions matching the updated app styling.
  • Open Settings directly from the sidebar gear and move documentation, diagnostics, onboarding, issue reporting and About into a separate question-mark help menu.
  • Move workspace tools into the Threads and Bots mode strip with matching icon buttons and a … menu to open tools or pin and unpin them per workspace, including when the sidebar is collapsed.
  • Redesign Add project as a compact local, remote, and new-project chooser, with matching dialog styling throughout project setup.
  • Redesign Open remote project with a compact connection form, saved-host tabs, clearer connection choices, and a dedicated setup view.
  • Move Connections, Home terminal and Keyboard shortcuts from the settings menu into sidebar footer buttons, with wrapping for narrow sidebars.
  • Move the theme toggle from the settings popout to a sun/moon button at the bottom of the sidebar, available when expanded or collapsed.
  • Match light-mode surfaces to a neutral palette with white content panels, near-white sidebars and a soft gray title bar and mode rail.
  • Match the app title bar to the Bots/Threads mode pane background in light and dark themes.
  • Soften light-mode borders across surfaces, menus, panels, buttons and form controls.
  • Give the Bots/Threads mode strip a slightly darker background than the sidebar in light and dark themes.
  • Enlarge the Threads and Bots icons and add more space on either side in the sidebar toggle.
  • Match main workspace tabs to the compact rounded terminal and browser tabs, including shaded selection and pill-style overflow controls.
  • Redesign the task center to match the new workspace with edge-to-edge searchable task lists, compact board controls, completed-task filters, and contextual thread actions.
  • Remove redundant workspace view headers so each view starts with its own tabs, controls, or heading.
  • Resume Codex sessions without transferring their full native transcript into Nora.
  • Show provider icons in thread settings provider and model dropdowns, including selected values and unavailable saved choices.
  • Keep workspace session tabs at a fixed width, with overflow tabs in a More dropdown and the selected tab kept visible.
  • Add saved thread sorting by recent activity, oldest first, or title in either direction, including grouped and resumable threads.
  • Add matching icons to every Show, Source and Group by option in thread view settings.
  • Show the Settled thread tick in green in light and dark themes.
  • Merge resumable chats into the regular thread list without CLI/Desktop badges, with saved per-app source filters, app logos, and optional source or provider grouping.
  • Show compact last-activity ages in thread rows, with exact dates on hover and live age updates.
  • Show the thread pin only while hovering or focusing a row, in place of its last-activity age so the row never shifts.
  • Pin threads started outside Nora straight from the thread list, importing them into Pinned without opening a tab.
  • Show the five most recent external threads by default, with Show more and Show less controls for older conversations.
  • Redesign workspace home with project and branch context, attention-first recent threads, live activity counts, and quick access to project tools.
  • Show provider icons beside threads in every grouping mode, keeping workspace avatars only in group headings.
  • Slightly darken the sidebar, title bar, and matching shell surfaces in dark mode.
  • Add slightly more left and right padding to the thread list.
  • Move the IDE shortcut and chooser from the title bar into floating thread details, opening the thread’s checkout.
  • Remove duplicate Files and Changes workspace views and shortcuts, keeping those tools in the right sidebar and Worktrees available separately.
  • Reorganize workspace tools into resumable work surfaces, consolidate terminals and worktree management, move connections into Settings, and remove the duplicate footer tools and AI chatbot shortcuts.
  • Consolidate thread scope into a compact header and filtering and grouping into one View options panel with an active indicator and reset control.
  • Scroll truncated thread names on hover or keyboard focus to reveal the full title, respecting reduced motion.
  • Combine provider, model, reasoning, and permissions in one picker with hidden scrollbars, clear explanations, and the selected mode and reasoning displayed beside the model in chat.
  • Include the selected workspace name in the new-thread greeting, updating it when switching workspaces.
  • Add dev:chat-simulator:nogpu and a --disable-gpu option for launching the chat simulator with hardware acceleration disabled.
  • Simplify Providers settings to installed providers with sign-in status and enable switches plus a default provider, with uninstalled providers on an Available to install tab; skills stay on the Skills page.
  • Limit supported providers to Claude Code, Codex and agents with an ACP runtime, removing Aider, Continue, Codebuff, Crush, Rovo Dev, Hermes, OpenClaw and the legacy Gemini CLI in favour of Google Antigravity.
  • Widen the floating thread details panel slightly while keeping the chat transcript and composer clear of it.
  • Represent skills with a document icon in thread details and Skills settings instead of the generic sparkles icon.
  • Show provider-reported usage reset times in the local timezone, with an explicit fallback when a reset time is unavailable.
  • Represent thread input, approval, failure, unconfirmed outcome, paused queue, and interrupted states with distinct coloured icons and tooltips in the shared indicator slot.
  • Keep sidebar threads on one line without chat icons, sharing a right-side slot for activity spinners and unread completion dots that clear when viewed.
  • Turn the composer Send button into Stop while work is running and the input is empty, retaining Queue for typed follow-ups and removing the separate interrupt/status row.
  • Reload simulator UI and styles automatically during development while preserving chats, drafts, theme, and running turns.
  • Stack queued messages in a narrow tab above the chat input, with a padded, shaded header, compact rows, inline editing, removal, and paused-queue controls.
  • Tighten corners across chat, pickers, sidebar controls, popovers, and Settings using shared 8px container and 5px control radii.
  • Present final turn changes in a compact file card with tighter corners, line totals, per-file diffs, and expandable file lists.
  • Pace the developer chat simulator with thinking delays, gradual text streaming, and visible tool execution instead of instant results.
  • Show live elapsed time for each running turn and retain its final duration in chat history, excluding time spent queued.
  • Group each turn's tool calls into one live activity line that expands to the full tool history, with failed actions visible in the summary.
  • Show a spinner beside sidebar threads while they are starting, working, or stopping, respecting reduced motion.
  • Show tool calls as compact, frameless single-line rows with live status, click-to-expand details, and explicit failure or unconfirmed outcomes.
  • Present action approvals with clear command, file or network context and descriptions of each permission scope.
  • Simplify navigation with a searchable workspace picker, saved role-based tool pins, and a unified thread list that opens workspace tools while preserving thread and commit drafts.
  • Rename the Agents settings section to Providers, including its default-provider controls and search entries.
  • Open and hide the local terminal from a sidebar icon, removing its persistent bottom strip while preserving the shell session.
  • Open Settings in a dismissible modal while keeping the active workspace and terminal mounted behind it.
  • Soften dark-mode surfaces with lighter charcoal shades, placing sidebars and headers above darker content and keeping panels, menus, and controls distinct.
  • Consolidate sidebar utilities into adjacent Usage, Workspace tools, Terminal, and App settings icons without a top divider, while removing the footer and keeping controls available in Settings.
  • Greet users by their first name in the configured accent color when starting a new chat, updating when their saved name changes.
  • Use native system fonts by default throughout Nora, with a System option in appearance settings.
  • Simplify work-role onboarding with a centered Nora logo, staggered entrance animations, compact choices, Skip, and seamless window chrome over a drifting blue glow, respecting reduced motion.
  • Prepare native Claude and Codex conversations with durable queues, checkout conflict checks, paginated history and recovery without repeating uncertain instructions.
  • Finalize native conversation reviews separately from execution, with explicit workspace scope and unavailable evidence after interrupted review.
  • Preserve before-and-after turn checkpoints and review diffs that exclude pre-existing edits, with explicit limits and recovery when attribution is unavailable.
  • Place the composer directly beneath the greeting on the new thread screen instead of at the bottom of the window.