Pick Your Failover & Per-Model Usage
Rate-limited? Choose from every available provider, not just the default — and see exactly which models are driving your usage. Plus a new Spaces mode.
featureimprovementfix
NucleusNucleus · Changelog
What's new, in print.
Every Nucleus release — features, fixes, and the small craft work in between. Written down so it doesn't disappear into a commit log.
Stability Improvements
Under-the-hood stability and reliability improvements.
improvement
Create Issues from Agent Plans
Turn an agent's plan into a tracked issue in one click, straight from the plan view.
featureimprovement
Org-Shared Skills
Skills published to your organization are now available to every agent, with trust gates applied consistently. And updates wait for your queued messages to finish before restarting.
featureimprovement
Terramind Tools & Redesigned Browser
A new set of built-in tools for video and audio understanding, image and social search, and live market data. Plus the browser panel gets a cleaner floating chrome that stays out of your way.
featurefix
Compute in Agent Mode
Recorded compute jobs now work in Agent mode, not just Science mode — so any agent can run reproducible work on your machine or a cloud GPU.
featureimprovement
Startup & Update Fixes
A proper loading screen on launch, plus further auto-updater and forecasting fixes.
improvementfix
Update Reliability
Fixes to the auto-updater and to resuming Grok sessions.
fix
Time Series Forecasting
Ask what happens next to any numeric series — sales, traffic, metrics, sensors — and get a forecast with honest uncertainty bands, running entirely on your own machine.
feature
Code Mode & Recorded Compute
New Code mode strips away everything but coding for a faster, focused agent. And compute jobs are now recorded and reproducible — run them locally or on cloud GPUs, with every run captured alongside its command and commit.
featureimprovementperformance
AI Teammates
Give your agents distinct identities and put them to work in their own sub-chats. Each teammate keeps its own persona and memory, so you're delegating to a specific colleague rather than a generic assistant.
featureimprovement
Better Deliverables for Non-Coding Work
Research, analysis and written work now come back as proper documents you can open and share, instead of walls of text in the chat.
improvement
Hit a Rate Limit? Keep Going.
When a provider rate-limits you, Nucleus now offers to continue on another one instead of stopping. Pick which model each fallback uses, or let it switch automatically every time.
featureimprovement
Smoother Long Conversations
Very long threads stay responsive regardless of which provider you're using.
performance
Rich Space Setup & Memory Improvements
Setting up a Space now populates your design system, skills and preferences automatically — and renders a live design showcase. Plus large chat histories use far less memory.
featureimprovementperformance
Separate Instruction Docs & Publishing to Spaces
Keep AGENTS.md and CLAUDE.md as distinct instruction sets, publish files from your project straight into your Space, and link projects that have no git remote.
featureimprovement
Your Agent Config, Shared With Your Team
Project instructions, preferences and memory now live in a shared Space instead of a file on your laptop. Everyone on the team gets the same agent context automatically — and you get a personal Space of your own alongside it.
feature
Stability Improvements
Under-the-hood stability and reliability improvements.
improvement
Mode Defaults
Agent, Plan and Science are the modes enabled out of the box. Design mode is now opt-in.
improvement
Design Mode & Science Mode
Two entirely new ways to work. Design mode ships 154 real design systems and a browser to explore them, so what your agent builds actually looks designed. Science mode brings research skills, literature search, and native molecule and equation rendering. Nucleus now speaks design and research as fluently as it speaks code.
featureimprovement
Effort Analytics & Project-Aware Agents
See where engineering effort actually went across goals and decisions, and agents now know which project they're working in from the first message.
featureimprovementfix
Linked Issues in Project Settings
See every issue linked to a repository, with its branch and pull request state, right in project settings.
featurefix
Model Picker Fixes
Claude Opus 4.8 no longer disappears from the picker when the model list can't load, and models show their real name instead of a generic label.
fiximprovement
Goals Tracking & Fullscreen Browser
Track your team's objectives with health signals and link decisions to the goals they serve. Plus a fullscreen browser panel and links that open in Nucleus instead of leaving the app.
featureimprovement
Chat Reload Fix
Fixed a crash when reopening chats containing certain tool results.
fix
File Watching Fix
Fixed a hang that could freeze the app when watching files in large projects.
fix
The Office Suite & The Embedded Browser
Nucleus is no longer just a code editor. A full Office suite — documents, spreadsheets, presentations, PDFs — now lives inside the app, and so does a real browser. Your agent can research on the web, build the deck, run the numbers, and write the doc. All in one place, all without ever leaving.
featureimprovementperformance
Shared Sign-In Across Nucleus
Sign in once and stay signed in across the desktop app and the command line — and signing out on one surface is respected everywhere.
improvement
Cleaner Reasoning Display
Reasoning blocks are merged into continuous sections instead of fragmenting into dozens of tiny pieces, and codebase indexing runs faster.
improvementperformance
Chat Timeline Minimap
Jump anywhere in a long conversation with a new timeline minimap that shows every turn at a glance.
feature
Grok Provider, Native Codex Protocol & Fast Reasoning
Grok is now available as an AI provider with Grok Build support, Codex adopts its native app server protocol for better reliability, and a new fast reasoning effort option speeds up quick tasks.
featureimprovement
Indexer Recovery & Performance
Stuck indexing operations now self-heal without needing a restart, and overall app performance is significantly improved.
bugfixperformance
Cloud Run Support & Auth Recovery
Chat sessions can now run in the cloud with Spaces sync, and the app automatically recovers from token expiry mid-conversation.
featureimprovement
Sync Indicator Fix
The sync indicator no longer gets stuck showing 'syncing' after an error.
bugfix
Sync Indicator Improvements
The sync indicator is now more accurate and recoverable when sync operations fail.
improvement
Smarter Rate Limiting & .env Protection
Background sync operations now use intelligent rate limiting to avoid overwhelming the server, and all .env file variants are properly excluded from indexing and git operations.
improvement
Agent Run Tracking & Spaces App Tools
Track detailed agent run history with a new activity dashboard, and agents can now interact with Spaces apps and database views directly.
featureimprovement
Agent Suggestions, Thinking Orb & Kanban Improvements
Agents can now proactively suggest work in a new Suggestions feature, a thinking orb shows live agent status, and the kanban board gains a needs-input column for conversations waiting on you.
featureimprovement
Claude Opus 5, Reasoning Visibility & Codex Catalog
Claude Opus 5 is now available, each message shows its reasoning effort level, and Codex dynamically discovers which models your subscription supports.
featureimprovement
Automation Management from Chat
Agents can now create, list, read, edit, and run automations directly from a chat session.
featureimprovement
Thinking Machines Support
Thinking Machines is now available as an AI provider alongside Claude, Codex, and Terramind.
feature
Parallel Code Reviews
Run multiple code reviews simultaneously across uncommitted changes, branches, and specific commits — all without interfering with each other.
feature
HTML Rendering in Chat & MCP Improvements
Agents can now render and preview HTML directly inside chat. MCP tools load faster and render more reliably across all providers.
featureimprovement
Spaces Integration & Indexing Status
Agents can now access your team's shared docs, channels, and calendar via the new Spaces MCP server. Codebase indexing status is visible in the sidebar and now works correctly in git worktrees.
featureperformance
Stability Improvements
Better handling of rate limits during indexing and sync, with automatic retries so background operations complete reliably.
improvement
Durable Chat Sync
Chat changes now sync reliably even if you go offline or the app closes unexpectedly — nothing gets lost.
improvement
Zero-Setup Codebase Indexing
Projects now connect for codebase indexing automatically when you send your first message — no manual configuration required.
improvement
Full Chat History & Agent Terminal Spawning
Load your entire conversation history without truncation, and agents can now spawn new terminal panes on demand.
feature
Annotate Code Selections & HTML Inspector
Add context notes to selected code before sending to the agent, inspect elements inside iframes in the HTML preview, and sub-chats now correctly fire their opening messages.
featurefix
Agent-Spawned Chats, Loops Mode & Insights
Agents can now spawn and message their own sub-chats, loops are a first-class chat mode with per-loop controls, and a new Insights dashboard shows your usage patterns and team analytics.
featureimprovement