# Claude 2.0: the Code tab and running work (/claude-and-chatgpt-2/tour/claude-code-tab) --- type: Document title: "Claude 2.0: the Code tab and running work" description: "The rebuilt Code tab, asking without interrupting, keeping work running with /goal and the cloud, running several things at once, projects, mods, and briefing Claude." status: stable order: 0.22 ksor: owner: team:panaversity audience: [ public ] approval: by: process:panaversity at: 2026-11-10T13:28:47Z chapter: "F00T" part: Front expert_status: required concepts: [ "Claude Code tab" ] last_verified: 2026-11-05 sources: - id: context title: "Use Claude Code Desktop (Anthropic, verified 2026-11-05)" resource: https://code.claude.com/docs/en/desktop - id: rewind title: "Claude 2.0 full guide: 12 changes in the new Claude desktop app (DEV Community, verified 2026-11-05)" resource: https://dev.to/hisashispace/claude-20-full-guide-12-changes-in-the-new-claude-desktop-app-and-the-ones-burning-your-usage-3nk4 - id: goal title: "Keep Claude working toward a goal (`/goal`) (Anthropic, verified 2026-11-05)" resource: https://code.claude.com/docs/en/goal - id: nested title: "Video walkthrough of the new Claude desktop app (YouTube, verified 2026-11-05)" resource: https://www.youtube.com/watch?v=tM0dwXhOv2o - id: coordinate title: "Use Claude Code Desktop (Anthropic, verified 2026-11-05)" resource: https://code.claude.com/docs/en/desktop - id: projects title: "Let Claude coordinate ongoing work with Projects (Anthropic, verified 2026-11-07)" resource: https://code.claude.com/docs/en/claude-projects.md - id: mods title: "Customize Claude Code with mods (Anthropic, verified 2026-11-05)" resource: https://claude.com/resources/articles/claude-code-mods - id: brief-claude title: "Prompting Claude Opus 5.5 (Anthropic, verified 2026-11-05)" resource: https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-opus-5-5 - id: cloud-claude title: "Claude Cowork architecture overview (Anthropic, verified 2026-11-05)" resource: https://support.claude.com/en/articles/14479288-claude-cowork-architecture-overview - id: claude-code-desktop title: "Use Claude Code Desktop (Anthropic, verified 2026-11-05)" resource: https://code.claude.com/docs/en/desktop - id: chatgpt-features-video title: "Video walkthrough of ChatGPT's newest features (YouTube, verified 2026-11-05)" resource: https://www.youtube.com/watch?v=9l9BVyCQPRk - id: anthropic-claude-code-cloud title: "Use Claude Code in the cloud (Anthropic, verified 2026-11-05)" resource: https://code.claude.com/docs/en/claude-code-on-the-web.md - id: projects-video title: "Video walkthrough of the new Claude app: projects, side chats, split view and artifacts (YouTube, verified 2026-11-07)" resource: https://www.youtube.com/watch?v=KjuANoI8HDw generated: at: 2026-11-10T13:28:47Z by: process:panaversity trust_tier: unverified build_id: sha256:46fd53c5d7ab1c14f6347908d45c9fdaecda4d914439923ee878bdcecd57803e dirty: true ksor_version: 0.0.60 --- ### The Code tab's new layout The Code tab was rebuilt so you can see what Claude is doing without reading its transcript. ![An illustrative layout of the Code tab, not an exact screenshot, in three columns. 1, Sessions: a pinned coordinator session, then sessions grouped into sections such as Client A and Book, with two nested sessions under Supplier research, marked as a reported feature. 2, Conversation: your brief with its outcome and "Done when" sentence, Claude's plan, progress and results of checks, your follow-up, the message box with a context ring, and buttons for the Terminal, Changes, Browser and Files panes. 3, Session rail: an Environment card showing local or cloud, the folder and a Remote Control switch, a Context card showing how messages, connectors, skills and memory fill the context, and a Changes card listing changed files. Captions: group work by project or client, open panes to inspect the work, and see where it runs, context use and changed files.](img/claude-code-tab-layout.png) **The session rail** is a column of cards beside the conversation, which you can show or hide. **Environment** shows where the session runs (on your computer, in the cloud, or on a remote machine), which folder it works in, and a switch for Remote Control. **Context** shows how full the session's context is: everything Claude holds in mind at once, meaning your messages plus the connectors, skills, agents and memory files it has loaded. When it fills, older detail is compacted into a summary, so the card helps you spot a connector or memory file crowding out your work. **Changes** lists what the session has changed, so you can open each change and comment on it.[^context] **Panes** open beside the conversation: a terminal, the list of changes, a browser that previews what Claude is building, the files it is working with, and its plan and background tasks. You may never open most of them. They are there for the moment you want to see for yourself. **Sections** group the sidebar's sessions by project, client or topic, under names you choose. A walkthrough shows how: open the **⋮** menu next to a session, move it to a new section, name the section, then drag other sessions into it *(reported)*.[^projects-video] A walkthrough also shows a **Connectors** card and an **Outputs** card, which lists every file the session made or used, and a button on the Context card that compacts the session: Claude summarizes the conversation so far to free up room *(reported)*. When Claude makes a plan, it appears as a to-do list in the conversation, and Claude ticks off each step as it goes *(reported)*.[^projects-video] Two settings help if you are not technical. In a local session you can open a file in the files pane and edit it yourself. Cloud sessions show files view-only, so ask Claude for the change. And **Ctrl+O** switches the conversation between three views: **Normal**, which folds each step into a short summary, **Thinking**, which adds Claude's thinking, and **Verbose**, which shows every step. Stay in Normal to see what Claude is doing without the technical detail.[^claude-code-desktop] ### Asking without interrupting Claude often works for minutes at a time, and you will have questions while it does. - **Side chat.** Open one with `/btw` or a keyboard shortcut. It can read the main conversation up to that point and answer you, but nothing you say there is added back. Side chats are not saved after you close the app. A walkthrough of the new app shows more than Anthropic's documentation describes: a side chat opened from the **⋮** menu at the top right, several side chats at once, side chats that are saved, and a side chat that sends one message back to the main conversation, such as only the fixes that matter from a review *(reported)*. Until Anthropic documents this, assume a side chat is not saved.[^projects-video] - **Fork.** Copy the conversation from a chosen point into a new session, to try a different direction without disturbing the original. - **Rewind** *(reported)*. Go back to an earlier point in the session with `/rewind`, like undo.[^rewind] - **Split view.** Open sessions side by side, from a session's **Open in** menu or by Option-clicking (Alt-clicking on Windows) it in the sidebar. A walkthrough also shows dragging a session from the sidebar into the one you are in *(reported)*.[^projects-video] ### Keeping the work running when you step away - **Remote Control** connects a session on your computer to the Claude mobile app and the web, so you can read it, answer its questions and approve its requests from your phone. Turn it on per session from the session rail, or in Settings turn on **Connect new sessions to Remote Control**. - **Keep computer awake** stops your computer from going to sleep from inactivity while the app runs your work. Closing the laptop lid still puts it to sleep. - **Cloud sessions** run on Anthropic's computers, so they continue with your laptop closed. A cloud session in the Code tab cannot see the files on your computer. It works from what you give it, such as a GitHub repository or uploaded files. - **Scheduled tasks** start a session at a set time. A local task needs your computer on. A cloud task does not, and can also start on a GitHub event or an API call. These pair well with **`/goal`**, which keeps Claude working until a condition you set is met. You state the finish line. In the weekly supplier payment run at Brightline Wholesale Supply, the company whose bill payments you follow through this book, it might be "all 15 invoices have a pay, hold or ask decision, and the run total matches the checking program". After each turn, a separate small model reads the conversation and judges whether the condition holds. If not, Claude starts another turn instead of stopping to ask you. The goal also ends if the checker judges the condition impossible, or if a turn fails on an error you have to fix, such as a used-up credit balance. You can clear a goal yourself at any time with `/goal clear`.[^goal] Two cautions. Write a finish line Claude's own output can show, never "make it great". And the separate model only reads what the conversation shows. It does not open files or rerun anything. It is a second opinion, not an independent check. A met goal is evidence that the work is finished, and it never makes the next action permitted. ![How /goal keeps Claude working: set a finish line, then work, check and continue. 1, Set the goal, such as all 15 invoices have a decision and the total matches the checking program. 2, Claude works one turn, runs the checks and shows the results. 3, The evaluator, a separate small model, reads the conversation and judges whether the evidence shows the goal is met. If not yet met, Claude takes another turn, guided by the reason. If met, the goal clears and you review the evidence. If judged impossible, the goal clears and you review the blocker. An error that needs you to fix it can also clear the goal. Two notes: the checker sees only the conversation, does not open files or rerun checks, and is a second opinion, not independent verification. And completion is not permission: a met goal is evidence of completion, and permissions and required approvals still control what happens next. At the foot: you can clear the goal yourself with /goal clear.](img/goal-loop.png) Read the figure from left to right, then follow the curved arrow back for each turn that falls short. On the right are the two ways the checker ends a goal: met, or judged impossible. The note in the middle is the third way, an error you must fix. The two notes along the bottom are the cautions above. ### Where the work actually runs It helps to know what happens behind the screen when Claude "keeps working in the cloud".[^cloud-claude],[^anthropic-claude-code-cloud] - **A fresh virtual machine for each session.** A cloud session runs on a virtual machine: a complete computer, simulated in software, on Anthropic's servers. Each session gets its own, created when the session starts and deleted when it ends. Sessions never share one, even inside the same company. - **Idle sessions lose their machine.** In the Code tab, if a cloud session sits idle long enough, Anthropic takes its virtual machine back. When you reopen the session, Claude starts a fresh one. Your conversation comes back. Work that was still running in the background does not, and Claude tells you what stopped so you can rerun it. - **What survives and what does not.** The conversation and the files saved to your account survive. Anything Claude installed or left on the virtual machine does not. So ask for results as files you keep. - **What it can reach.** Only websites on an allowed list, through a filter the session cannot change, and nothing inside your company's own network. Enterprise organizations start with no internet access at all. Your keys for GitHub and your connectors stay outside the machine, and Anthropic's servers add them when a request leaves it. - **Your own files.** A cloud session reaches your computer only through the desktop app, and only folders you have connected. Each request is checked against your permissions. Files read this way are processed on Anthropic's servers, not only on your computer. Why does it work this way, and what does it mean for an AI Worker? See [Behind the Screen](../behind-the-screen/why-the-borrowed-desk-gets-cleared.md). - **Local sessions.** A Code tab session set to Local runs on your own computer, in the folder you chose. Older tasks started on the desktop before October 2026 ran inside a virtual machine on your own computer instead. Anthropic documents a new machine per session. It does not say whether each new task inside the same conversation starts a new session, so do not count on anything carrying over between tasks except your conversation and your files. ### Three ways to run several things at once **Nested sessions** *(reported)* split one job into side jobs. They are described in walkthroughs of the new app but were not yet in Anthropic's documentation when this page was verified. Claude can start sessions of its own that run in parallel, appear under the main session, and report back to it. Each is a full session you can open and talk to directly: you might have five competitors researched at once, then ask about one of them without cluttering the main conversation. Ask for it ("split this into a separate session for each competitor"), or start one from a card in which Claude suggests related work.[^nested],[^chatgpt-features-video],[^claude-code-desktop] On a Team plan, where projects are not yet offered, a walkthrough reports nested sessions as the same idea in another form: each side session gets a brief from the main session instead of shared memory, and it runs on your computer instead of in the cloud *(reported)*.[^projects-video] **A coordinating session** manages sessions already running. This is a way of working, not a button. Every Code session can list, read, message, rename and archive your other sessions, so you pin one to the top of the sidebar and use it as a control center. Ask it "what's waiting on me?" and it tells you which sessions have finished, which need approval and which are still working. It sees your 20 most recently active sessions on this computer, not cloud sessions, unless you ask it to look further.[^coordinate] **Projects** carry ongoing work with shared instructions and memory. A project is one ongoing conversation in which Claude acts as a coordinator: you tell it what needs doing, and it starts a thread for each task. Threads run in the cloud by default, so they keep going with your laptop closed. The next section covers projects in full.[^projects] ![Three ways to run several things at once: split one job, coordinate separate jobs, sustain ongoing work. 1, Nested sessions, a reported feature not yet documented: a main session assigns side jobs 1, 2 and 3, and each side job reports back. You can open a side job and ask it directly. 2, A coordinating session: a pinned coordinator reads, messages and organizes sessions already running, such as Build, Research and Report. Ask it "What's waiting on me?" for one place to review ongoing sessions. 3, A project: one project conversation with shared instructions and memory starts cloud threads. Cloud work can continue with your laptop closed, and the overview shows what needs you. Local resources may still need your computer online. A band says more parallel work uses your plan's limits faster, so start with two sessions.](img/three-ways-parallel.png) The arrows show who starts the work. In nested sessions the main session starts its side jobs and they report back. A coordinator starts nothing. It watches and steers sessions that already exist. A project starts cloud threads that share its instructions and memory. | You want to | Use | Context each piece starts with | Runs with your laptop closed | | --- | --- | --- | --- | | Split one big job into parallel parts | Nested sessions | What the main session tells it | Only if it runs in the cloud | | Keep track of sessions already running | A coordinating session | Your other sessions, as it reads them | No | | Run an ongoing stream of related work | A project | The project's instructions and, in the cloud, its memory | Yes, by default | Parallel work uses your plan's limits faster, and so does standing context: connectors, skills and memory files load into every new session, and each side-chat question rereads the main conversation. Start with two sessions and add more once you trust the results. ### Projects: one conversation that runs the work Before, a project was a folder of conversations that could not see each other. On a big job, you ran five conversations yourself and pasted the same background into each one. A rebuilt project does that coordinating for you. You talk to it the way you would brief a team lead.[^projects] **How work flows.** Suppose you send one message: "Check this week's 15 supplier invoices, draft the payment summary for the owner, and update the supplier contact list." Claude starts three threads, one for each job, and each appears as a card under your message. A thread is a full session with its own context. It shows its plan as it works and reports back to the project conversation when it finishes. A quick question gets an answer in place, and a follow-up goes to the thread already working in that area. If you ask for something unrelated in the middle of a thread, Claude may offer to start a new thread for it *(reported)*. That is a good habit anyway: one thread per job.[^projects-video] **Steering a thread.** Click a thread's card to read what it did, write to it directly, or stop it. A permission request waits inside its own thread. Telling Claude in the project conversation to go ahead does not reach it, so answer it in the thread. **The project conversation stays clear.** It sees what threads report, not every step they take. It works from recent messages, recent threads and the project's memory, not its whole history, so it stays sharp for as long as the project runs. Ask it "where are we with everything?" and it answers without you opening a single thread. **Context goes in once.** - **Project instructions** are the brief every new thread starts from, up to 16,000 characters: what the project is for, how a thread checks its own work, what to do when something is missing, and what needs your approval. - **Project memory** holds the requirements, decisions and pitfalls Claude writes down as it works. When you correct one thread, tell Claude to remember the correction, and every later cloud thread starts with it. - **The Library**, a tab in the **Overview** pane, holds every file you added and every file the threads made, so nothing is lost across many conversations. The pane's **Threads** tab sorts threads by state, such as **Waiting on you**, **Working** and **Ready for review**. **Cloud by default, your computer when needed.** Cloud threads keep going after you close your laptop, and you can check on them and steer them from your phone. When a job needs files or tools that are only on your computer, choose **Work locally** from the **+** menu beside the message box, then allow the folder. That thread runs on your computer through Remote Control, and only while the computer is awake. It starts with the project's instructions but not its memory. Check four things before you start a project: 1. **Model and effort.** Every thread starts on Opus at high effort, which uses your plan's limits fastest. Open **Project settings > General** and check **Thread model** and **Thread effort** first. 2. **Connectors.** Cloud threads can use the connectors on your claude.ai account. The project conversation itself has none, so it hands any work that needs a connector to a thread. 3. **Skills and settings.** Cloud threads do not load the skills, plugins, connectors or settings on your own computer. Turn the skills you need on in your claude.ai account, and add plugins in **Project settings > Plugins**. 4. **Your plan.** Projects are in beta on Pro and Max plans, and a project belongs to one person. You cannot share it or its threads. Use a project when a goal outlasts one session and keeps producing tasks, such as a weekly payment run with its follow-up questions. For one task that fits in one session, start a normal session. For work where every task needs your computer, use local sessions. ### Mods: changing the app itself Mods change how Claude Code looks and behaves. Describe the change, for example "make me a mod that shows my weekly usage above the message box", and Claude writes the mod, installs it, and reloads it in your session. A mod can add a pane or a status line, show a notification, rewrite or block a tool call, answer a permission request, or hide secrets from the screen. Anthropic's examples include a pane that follows your builds, a confirmation step before any command touches production settings, and an audit log of what other mods do.[^mods] Mods are not sandboxed: they have the same access to your computer as Claude Code itself. Install only mods you have read or that come from a source you trust. In companies, administrators decide which plugin sources are allowed, and a built-in safety mod loads first to block the riskiest changes. ### Briefing Claude 2.0 The new app needs less instruction than most people still give it. Anthropic's guidance for Opus 5.5 is to start at medium effort, the default, and test other levels on your own work. It suggests stating the condition for done up front, and removing instructions that tell Claude to think carefully, because the model decides how much to think and the effort setting is the main control.[^brief-claude] Panaversity's own research adds two habits: give the whole task in one message, and leave out lines such as "double-check everything". A specific condition for done adds far more. Chapter 5 teaches this as the [Four-Part Brief](../../managing-ai-workers/the-four-part-brief/overview.md): the **outcome** you want, ending with a "Done when" sentence, the **context** (who it is for and what to work from), the **format** of the result, and the **constraints**, the rules the result must follow and how far the worker may go alone. The payment-run brief on the first page shows all four. The new apps shift its emphasis. The outcome's "Done when" sentence is one Claude can check itself. Every brief still covers all four parts, but context, format and constraints can be short when something else already settles them, such as project instructions, approved sources and enforced permissions. End with "Show what you checked and what the checks found, and list anything you could not verify", so that you review evidence, not a claim. When you are not sure what you want, ask Claude to interview you until it could write the brief itself, then start a fresh session from that brief. [^context]: Type `/context` for a full breakdown of what fills the context. A small context ring next to the model picker shows the same information at a glance. Use Claude Code Desktop, Anthropic. [^rewind]: Claude 2.0 full guide: 12 changes in the new Claude desktop app, DEV Community. Not confirmed on a first-party page when this page was verified. [^goal]: `/goal` works in the Code tab, in the terminal and through Remote Control. The evaluator is the small, fast model configured for your account, Haiku by default. Keep Claude working toward a goal (`/goal`), Anthropic. [^cloud-claude]: As verified on November 5, 2026. Anthropic: "Each session gets its own sandbox, created when the session starts and destroyed when it ends," and sandboxes "don't share state with each other or across organizations." The Code tab's cloud sessions run in "an isolated, Anthropic-managed VM", which is "reclaimed" after a period of inactivity. Local Cowork sessions, which ran code in "a dedicated Linux VM" on the computer, apply on Pro, Max and Ultra only to tasks started before October 6, 2026. Claude Cowork architecture overview, Anthropic. [^anthropic-claude-code-cloud]: Use Claude Code in the cloud, Anthropic. [^nested]: Nested sessions are described in the video walkthroughs and in reporting on the desktop app. The desktop documentation describes the cards that suggest related work, which start it in another session, but nested sessions themselves were not confirmed on a first-party page when this page was verified. [^chatgpt-features-video]: Video walkthrough of ChatGPT's newest features, YouTube. [^claude-code-desktop]: Use Claude Code Desktop, Anthropic. [^coordinate]: Claude asks before archiving a session, except in Auto and Bypass permissions modes. Messages to a busy session are held until it finishes its current work. Use Claude Code Desktop, Anthropic. [^projects]: As verified on November 7, 2026. Projects were in public beta on Pro and Max plans and rolling out gradually, and were not yet offered on Team or Enterprise. They work at claude.ai/code, in the Code tab of the desktop app and in the Claude mobile app, but not in the terminal or the code editor extensions. Projects in conversations keep working as before until the new kind reaches them. The author confirmed in November 2026 that the new kind also works in conversations. To run a thread on your computer, that computer needs Claude Code v2.1.280 or later and, in the desktop app, **Use this computer from your phone and claude.ai** turned on under **Settings > Claude Code**, with the folder added. The enforced limit is 200 new threads a day across your projects. Let Claude coordinate ongoing work with Projects, Anthropic. [^projects-video]: Video walkthrough of the new Claude app: projects, side chats, split view and artifacts. Not confirmed on a first-party page when this page was verified on November 7, 2026. [^mods]: Mods became available in the Claude Code terminal and desktop app on October 1, 2026. They ship inside plugins. Customize Claude Code with mods, Anthropic. [^brief-claude]: As verified on November 5, 2026, Anthropic: "Start at `medium`, the default on Claude Opus 5.5," and "test several levels against your own evals." It also suggests stating "the completion condition up front". Prompting Claude Opus 5.5, Anthropic. The two habits are from Panaversity's research paper "Less Prompting, Better Briefs" (November 3, 2026). Osmani's article could not be opened.