More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Claude Office: Every Claude Code Chat on One Screen

    1

    A local dashboard that shows each Claude Code chat as a desk: working, stuck, or waiting for you. Answer Claude's questions with one tap. Runs on 127.0.0.1, standard-library Python, no API key.

    Free

    Security scanned
    Claude Office: Every Claude Code Chat on One Screen

    Claude Office: Every Claude Code Chat on One Screen

    Example session with this skill installed

    Set up Claude Office for me. I run Claude Code chats for three projects: my store in C:\Users\dana\Projects\shop, my blog in C:\Users\dana\Projects\blog, and client work in C:\Users\dana\Projects\clients.

    • Read your context and instructions
    • Compiled the claude office
    • Generated the document

    Claude Office is running

    Open http://127.0.0.1:8765. Your current chat is already on it as a green desk.

    What I did

    1. Folder. Created C:\Users\dana\claude-office (outside your projects, so nothing mixes with your code).

    Files. Copied the 8 office files: server.py, index.html, start.bat, start.command, office.example.json, tasks.example.json, VERSION, CLAUDE.md.
    3. Settings. Wrote office.json with your three projects:

    idName on the pageFolderMatches requests with
    shopStoreC:\Users\dana\Projects\shopstore, shop, product, order, checkout
    blogBlogC:\Users\dana\Projects\blogblog, post, article, newsletter
    clientsClient workC:\Users\dana\Projects\clientsclient, invoice, proposal, meeting

    Started it with python server.py 8765. The black terminal window is the engine; keep it open while you use the office.
    5. Checked it. The page loaded and this chat appeared as a desk within 10 seconds.

    What you'll see

    • One desk per chat from the last 24 hours. Green is working, orange stopped, blue finished and waiting for you.
    • The

    Ask box at the top: type "fix the checkout button on the store" and it opens a new chat in the shop folder with your words ready. You press Send.

    • The waiting on you strip stays empty until a chat leaves you a question.

    Next steps

    • Want Claude to leave you yes/no questions on the page? I can add a short section to your global instructions. I'll only do it if you say yes.
    • To start it again tomorrow: double-click start.bat in C:\Users\dana\claude-office.
    • Something looks wrong? Say "office feedback" and I'll write a ready review for this skill's page.

    index.html

    PDF · document

    Generated

    Example file from a real run - the skill writes it into your workspace.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Running three or four Claude Code chats at once? One is writing, one is waiting for your OK, one finished an hour ago and nobody noticed.

    Claude Office reads your own chat files and draws one desk per chat from the last 24 hours:

    • 🟢 working, 🟠 stopped (take a look), 🔵 finished and waiting for you, ⚪ on a break
    • a "waiting on you" card: Claude leaves a yes/no question with its recommendation, and you answer with one tap (Yes, No, Later)
    • an Ask box: write a request in plain words, the page picks the project and opens a Claude chat with your text already typed. You press Send.
    • Hebrew and English, 6 skins including dark mode, refreshes every 10 seconds

    How it works

    Your agent copies a small Python server (standard library only) and one HTML page into a folder you choose, asks which projects you have and where they live, writes the settings file and starts the page on 127.0.0.1:8765. It checks that your current chat shows up as a desk before it says it's done.

    Before / after

    BeforeAfter
    Five chat tabs. You click through each one to find the chat that stopped.One page, five desks. The one waiting for you is blue.
    Claude asks a yes/no question inside chat number four and waits there all afternoon.The question sits at the top of the office as a card, with Claude's pick and three buttons.
    A new request means finding the right project folder and opening a chat there.You type it in the Ask box. The page picks the project and opens the chat with your words ready.

    Safety

    • Runs on 127.0.0.1 only. No API key, no AI calls in the background, no outside servers.
    • Nothing is ever sent for you: every request becomes a claude:// link that you open and send yourself.
    • Emails, phone numbers and keys are masked before any line of a chat shows on the page.
    • Writes only inside the office folder.

    Known limitations

    • Needs the Claude desktop app with Claude Code, and Python 3.8 or newer.
    • Built and used daily on Windows. macOS should work through start.command but hasn't been tested on a real Mac yet.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.

    Requires Python 3.8+ and Claude desktop app. Primary support for Windows; macOS is less tested. Local host only.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 10 days ago

    • Free with an Agensi account

    Software
    Claude desktop app (Claude Code)Python 3.8+

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Network Access

    Reads your own Claude Code chat files (last lines only) and the desktop app's session list to draw the desks. Writes only inside the office folder (office.json, tasks.json, state/). The server binds to 127.0.0.1 and makes no outside network calls.

    Listed10 days ago

    What's inside

    Frequently Asked Questions

    Claude Office provides a persistent visual HUD that tracks multiple concurrent Claude Code sessions, highlighting which projects are active and which are waiting for your input so you don't have to cycle through terminal windows.

    The skill requires a local Python 3.8+ environment and the Claude desktop application. While it is optimized for Windows, it can run on macOS via the included script, though macOS support is currently in a testing phase.

    A standard-library Python server, one HTML page with six skins including dark mode, start scripts for Windows and macOS, example settings files and a CLAUDE.md with setup and feedback steps. Emails, phone numbers and keys are masked before any chat line shows on the page.

    No. Your agent does the setup: it copies the files to a folder you pick, asks which projects you have and where they live, writes the settings file, starts the page and checks that your chat shows up. You answer a few questions; to start it again later you double-click start.bat.

    Claude Office reads local session files and transcripts directly from your machine and processes them locally, ensuring your project data and code never leave your computer for the sake of the dashboard.

    New in Agents & Orchestration

    Free