Skip to content
Back to search

myoffice

SkillWorks with Claude CodeGitHub

Open a live office view of every Claude Code, Codex, Gemini CLI, Antigravity CLI, Copilot CLI, OMP, Pi, Kimi Code, DeepSeek Harness, Grok Build, OpenCode and Hermes agent on this machine — who is thinking, running a tool, writing a reply, waiting for the user, or done. Use when the user types /myoffice, asks what their agents are doing, wants to see all running sessions, or which agent is stuck or finished ("我的 agent 在幹嘛", "show my agents").

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

Its scripts were only partly scanned.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/Charonyuu/whatsmyagentdoing with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/Charonyuu/whatsmyagentdoing'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/charonyuu--whatsmyagentdoing/badge.svg)](/listings/charonyuu--whatsmyagentdoing)

HTML

<a href="/listings/charonyuu--whatsmyagentdoing"><img src="/api/listings/charonyuu--whatsmyagentdoing/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

**Easiest: ask your agent.** Paste this into Claude Code or Codex:

> Install the myoffice skill from https://github.com/Charonyuu/whatsmyagentdoing: clone it with `git clone --depth 1` into `~/.claude/skills/myoffice` (for Codex: `~/.codex/skills/myoffice`), check that `node --version` is 18 or newer, then tell me to start a new session and type /myoffice.

Or do it yourself:

# Claude Code
git clone --depth 1 https://github.com/Charonyuu/whatsmyagentdoing ~/.claude/skills/myoffice

# Codex
git clone --depth 1 https://github.com/Charonyuu/whatsmyagentdoing ~/.codex/skills/myoffice

Then type **`/myoffice`** (or just ask *"what are my agents doing?"*). The office opens in your browser.

Needs Node.js 18+ (22.13+ to include OpenCode and Hermes). Nothing to `npm install` — the repo runs as cloned. Prefer no skill? `node bin/cli.js` from a clone works too.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

SKILL.md · Manifest · 69 lines
  1. ---
  2. name: myoffice
  3. description: Open a live office view of every Claude Code, Codex, Gemini CLI, Antigravity CLI, Copilot CLI, OMP, Pi, Kimi Code, DeepSeek Harness, Grok Build, OpenCode and Hermes agent on this machine — who is thinking, running a tool, writing a reply, waiting for the user, or done. Use when the user types /myoffice, asks what their agents are doing, wants to see all running sessions, or which agent is stuck or finished ("我的 agent 在幹嘛", "show my agents").
  4. ---
  5. # myoffice
  6. Start the dashboard. It serves a local page and opens it in the browser.
  7. Run it **in the background** so it keeps running and doesn't block the chat
  8. (this skill's base directory is shown above as "Base directory for this skill"):
  9. ```bash
  10. node "<base directory>/bin/cli.js"
  11. ```
  12. If it is already running, the same command just reopens the browser tab.
  13. To **update** it — `/myoffice update`, or the user asks to update / upgrade —
  14. run `node "<base directory>/bin/cli.js" --update` (not in the background),
  15. report the version it printed, then start it again as above. The page also
  16. updates itself when a new version is out, and shows an "Update" button if that fails.
  17. To **close** it — `/myoffice stop`, or the user asks to stop / close / shut the
  18. office down — run (not in the background):
  19. ```bash
  20. node "<base directory>/bin/cli.js" --stop
  21. ```
  22. and say it's stopped.
  23. Then tell the user, in one line, that the office is open at the printed URL.
  24. If the output also says **"Update available: vA → vB"**, tell the user a new
  25. version is out; the page installs it on its own once it opens (or run
  26. `/myoffice update`, below).
  27. Requires Node.js 18+; if `node` is missing, say so and stop.
  28. ## Scenes (3D)
  29. Besides the built-in office there are downloadable 3D scenes (e.g. a pirate
  30. cove). When the user asks for another scene, a different look, "DLC", or names
  31. one ("海盜船", "pirate"):
  32. 1. `node "<base directory>/bin/cli.js" --scenes` — lists every scene, whether
  33. it's installed, and its download size. Use the ids it prints; don't guess.
  34. 2. `node "<base directory>/bin/cli.js" --scene <id>` (in the background) —
  35. downloads it if needed (checksum-verified, data only) and opens the office
  36. in that scene. `--scene office` goes back to the office.
  37. Scenes are only downloaded when asked for. `--remove-scene <id>` deletes one.
  38. Options the user may ask for:
  39. - `--private` — hide prompts/replies, show status only (for screen sharing)
  40. - `--hours 4` — only sessions active in the last 4 hours (default 24)
  41. - `--port 5000` — different port (default 4545)
  42. - `--stop` — stop the running dashboard
  43. - `--json` — print a one-off snapshot instead of opening a page. Use this when
  44. the user asks in chat what their other agents are doing: run
  45. `node "<base directory>/bin/cli.js" --json --hours 2` and summarize by status.
  46. If the user wants Claude Code permission prompts to show as "Waiting for you"
  47. (Claude Code doesn't log them), offer `node "<base directory>/bin/cli.js" --install-hook`.
  48. It adds one Notification hook to their Claude Code settings; `--uninstall-hook` removes it.
  49. Ask before running it, since it changes their settings.
  50. Everything is read locally from the agents' own session folders (`~/.claude/projects`,
  51. `~/.codex/sessions`, `~/.gemini/tmp`, `~/.gemini/antigravity-cli`, `~/.copilot/session-state`, `~/.omp`, `~/.pi`, `~/.kimi-code`,
  52. `~/.dsh`, `~/.grok`) and databases (`~/.local/share/opencode/opencode.db`, `%LOCALAPPDATA%\hermes\state.db`);
  53. the server listens on 127.0.0.1 only and sends nothing anywhere.
README.md · README · 127 lines
  1. <img src="assets/logo.svg" width="64" height="64" alt="">
  2. # whatsmyagentdoing
  3. **Type `/myoffice` and see what every coding agent on your machine is doing — live, in a little office.**
  4. Who is thinking, who is running a tool, who is waiting for you, who just finished. Works with Claude Code, Codex, Gemini CLI, Antigravity CLI, Copilot CLI, OMP, Pi, Kimi Code, DeepSeek Harness, Grok Build, OpenCode and Hermes. Nothing leaves your machine.
  5. **Website:** https://charonyuu.github.io/whatsmyagentdoing/
  6. https://github.com/user-attachments/assets/d29fac22-a4d8-4cca-8252-6fd34c7ce713
  7. | Click an agent to see what it's doing | Today's turns and tokens |
  8. |---|---|
  9. | ![Agent detail: status, model, folder, context and recent conversation](docs/detail.jpg) | ![Today panel: turns and tokens by agent and project](docs/today.jpg) |
  10. <sub>Recorded with `--demo` (a made-up team), so no real sessions are shown.</sub>
  11. ## Install
  12. **Easiest: ask your agent.** Paste this into Claude Code or Codex:
  13. > Install the myoffice skill from https://github.com/Charonyuu/whatsmyagentdoing: clone it with `git clone --depth 1` into `~/.claude/skills/myoffice` (for Codex: `~/.codex/skills/myoffice`), check that `node --version` is 18 or newer, then tell me to start a new session and type /myoffice.
  14. Or do it yourself:
  15. ```bash
  16. # Claude Code
  17. git clone --depth 1 https://github.com/Charonyuu/whatsmyagentdoing ~/.claude/skills/myoffice
  18. # Codex
  19. git clone --depth 1 https://github.com/Charonyuu/whatsmyagentdoing ~/.codex/skills/myoffice
  20. ```
  21. Then type **`/myoffice`** (or just ask *"what are my agents doing?"*). The office opens in your browser.
  22. Needs Node.js 18+ (22.13+ to include OpenCode and Hermes). Nothing to `npm install` — the repo runs as cloned. Prefer no skill? `node bin/cli.js` from a clone works too.
  23. ## What you get
  24. **The office.** Agents that are working sit at their desks: a thought bubble while thinking, code scrolling on the monitor while a tool runs, standing up and waving when they need you. Recently finished ones go off duty and wander — coffee in the pantry, the couch, ping-pong, a nap at an empty desk. Hover anyone to see what they're on; click to open them in the side panel.
  25. **Live status for every session:**
  26. | Status | Meaning |
  27. |---|---|
  28. | Thinking | reasoning, or reading a tool result |
  29. | Running | a tool call is in flight — shows which (`Bash: npm test`, `Edit app.ts`, `github · create_pr`) |
  30. | Writing reply | started its answer, turn not finished yet |
  31. | Waiting for you | asked you a question or needs approval |
  32. | Done | turn finished, with the last reply |
  33. | Interrupted / Error / Idle | |
  34. **The side panel** groups sessions into **Needs you → Working → Finished**. Click one to see its model, folder, today's usage, how full its context window is, a copy button for the resume command, and its recent conversation (your messages, its replies and the tools it ran). Session titles also float above the agents in the office; turn them off in **Settings → Session titles**.
  35. **Alerts.** When an agent needs you, the page can send a browser notification and play a short chime (**Settings → Sound**). A tool that runs for more than 10 minutes, or thinking that goes silent for 8, is flagged as *No progress*, so a hung command doesn't cost you an hour.
  36. **Today.** Turns and tokens since local midnight, by agent and by project.
  37. **2D or 3D.** **2D** is a hand-drawn canvas with no dependencies. **3D** adds real-time lighting and shadows using [Kenney](https://kenney.nl)'s CC0 models; drag to rotate, scroll to zoom. The 3D bundle only loads when you pick it.
  38. **Scenes (3D).** Besides the office there are seven downloadable scenes: Christmas Cabin, Neon Arcade, Haunted Hollow, Pirate Cove, Café, Space Station and Castle (about 100–290 KB each). Pick one from **Settings**, run `--scene pirate`, or ask your agent. Packs are data only and are checked against a pinned SHA-256 before they're installed.
  39. ![2D office with the agent panel](docs/office.jpg)
  40. **Pop out.** In browsers with [Document Picture-in-Picture](https://developer.chrome.com/docs/web-platform/document-picture-in-picture), the office or a compact list floats in an always-on-top window that stays in sync with the main tab. Elsewhere, **Pop out** starts a local, view-only video of the office you can float with the browser's own video PiP; no screen sharing, nothing uploaded.
  41. **Ten languages.** English, 繁體中文, 简体中文, 日本語, 한국어, Español, Français, Deutsch, Português and Русский. It follows the language you write your prompts in, falling back to your browser's; pick one from the toolbar to override.
  42. ## Supported agents
  43. | Agent | Read from | Resume |
  44. |---|---|---|
  45. | Claude Code (CLI, desktop, IDE) | `~/.claude/projects/` | `claude --resume` |
  46. | Codex (CLI, desktop, IDE) | `~/.codex/sessions/` | `codex resume` |
  47. | Gemini CLI | `~/.gemini/tmp/` | `gemini --resume` |
  48. | Antigravity CLI | `~/.gemini/antigravity-cli/` | `agy --conversation` |
  49. | Copilot CLI | `~/.copilot/session-state/` | `copilot --resume` |
  50. | OMP (oh-my-pi) | `~/.omp/agent/sessions/` — [details](docs/omp.md) | from its log |
  51. | Pi | `~/.pi/agent/sessions/` | `pi --session` |
  52. | Kimi Code | `~/.kimi-code/sessions/` | `kimi --session` |
  53. | DeepSeek Harness (DSH) | `~/.dsh/sessions/` | `dsh tui --resume` |
  54. | Grok Build | `~/.grok/sessions/` | `grok --resume` |
  55. | OpenCode | `~/.local/share/opencode/opencode.db` | `opencode --session` |
  56. | Hermes | `~/.hermes/state.db` (Windows: `%LOCALAPPDATA%\hermes`) | `hermes --resume` |
  57. All of them are read-only. Paths, environment overrides and per-agent details: [How each agent is read](docs/agents.md). Sessions seen by a local Magpie gateway can be added too: [Magpie](docs/magpie.md). The Google Antigravity desktop app keeps its conversations in an undocumented binary store and isn't supported yet; its `agy` CLI is.
  58. ## Options
  59. ```
  60. -p, --port <n> Port (default 4545; next free one if taken)
  61. --hours <n> Sessions active in the last n hours (default 24)
  62. --omp-dir <path> OMP sessions root, or a directory passed to omp --session-dir
  63. --no-open Don't open the browser
  64. --private Hide prompts, replies and titles — status only (screen-share safe)
  65. --json Print one snapshot as JSON and exit
  66. --demo Show a made-up team instead of your sessions (screenshots, trying it out)
  67. --stop Stop a running dashboard (or say /myoffice stop)
  68. --update git pull the latest version and restart (or say /myoffice update)
  69. --scenes List 3D scenes, installed and downloadable
  70. --scene <id> Download a scene if needed and open the office in it
  71. --install-hook Optional: add a Claude Code hook so permission prompts show "Waiting for you"
  72. --uninstall-hook Remove that hook again
  73. ```
  74. Running it again while it's up just reopens the tab.
  75. **Updates:** it keeps itself up to date. The page checks this repo's `package.json` about every half hour; when a newer version is out it runs `git pull --ff-only` in the install folder and restarts the dashboard. If that pull fails (say you changed files in the clone), an **Update** button stays on the page instead. `--update` or `/myoffice update` does the same from the command line. Set `WMAD_NO_UPDATE_CHECK=1` to turn checking and updating off.
  76. ## Privacy
  77. Nothing leaves your machine. The server binds to `127.0.0.1` only and rejects requests that aren't addressed to localhost. It tails the agents' own log files read-only and pushes statuses to the page over Server-Sent Events. No runtime dependencies — read `src/` in five minutes. With `--private`, prompts, replies and titles never reach the page.
  78. ## Limits
  79. - Logs are written per finished block, not per token, so "Writing reply" means a reply block has landed — not a live keystroke stream.
  80. - Claude Code doesn't log permission prompts, so a tool waiting on your approval shows as "Running" until you answer. `node bin/cli.js --install-hook` adds an optional hook that fixes this ([details](docs/agents.md#agent-specific-limits)).
  81. - The log formats aren't public APIs. If an update changes them, statuses may degrade — please open an issue.
  82. ## Contributing
  83. PRs are welcome — several agents and languages here came from the community. See [Development checks](docs/development.md) for the test commands.
  84. ## License
  85. MIT. 3D models in `assets/kenney/` are by [Kenney](https://kenney.nl), released under CC0 (see the license files there).

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.