Agent Standup

See who is waiting on whom, and who is blocked — across all your Claude Code, Codex and Mistral Vibe sessions — at zero token cost. Agent Standup turns the session files Claude Code, Codex and Mistral Vibe already write into a live team call (and a board of every session): it only reads local files and never calls Claude unless you turn on French translation or the jokes.

Your Claude Code, Codex or Mistral Vibe session as a live team call. The lead gives orders, sub-agents join the call, work at their desks, share their screen and report back — with voices, expressions and a view of the real work as it happens.

Runs as a VS Code extension, a terminal command, or a desktop app. Try it in 30 seconds, no install and no editor (Node 18+): npx @double_geste/agent-standup --open, then ⚙ → Source → Demo.

Works with: Claude Code · Codex · Mistral Vibe — in VS Code, Cursor, Windsurf, any terminal, or as a desktop app.

Just installed the extension? Click the Agent Standup icon in the Activity Bar (or run Agent Standup: Open live call), then press Join. No session yet: press ▶ Watch the demo.

→ Website & download: https://agent-standup.com

🌐 Français · 繁體中文 · 简体中文 · Español · Deutsch · Dansk · Português · 日本語 · 한국어

A fictional team fixing a slow login page — the built-in demo (⚙ → Source → Demo). Watch the full 1-minute video.

New: Agent Standup is on YouTube. The first video is the very session that made it, replayed in the app with its own voices (2 min 51). Subscribe to the channel.

Halloween pack (optional)

Eight characters drawn by hand by Fabien Coupas, each in its own room: a pumpkin leads the call, you are a vampire, and a ghost, a cat, a bat, a wolf, a witch and a crowd of skeletons do the work.

It is entirely optional. Agent Standup is and stays free and MIT licensed, and nothing in it is locked behind the pack: every feature and the default characters stay as they are, and the vampire is free for everyone. Buying the pack (2 €, or more if you like) is simply a way to support the artist who drew it and the developer who makes the app; the money is shared between them.

See the pack, or in the app: ⚙ → Characters → Halloween pack. The pack's pictures are not part of the open-source code: they remain the artist's work.

Unofficial community project, not affiliated with or endorsed by Anthropic, OpenAI or Mistral AI. "Claude" is a trademark of Anthropic; "Codex" is a trademark of OpenAI; "Mistral" is a trademark of Mistral AI.

What you get

• Token usage per agent — a ⚡ counter in the header, on each tile, on each line of the agents list and in the camera of the agent who shares its screen shows how many tokens each agent used (input + output + cache writes; cache reads shown separately in the tooltip), counted once per message as recorded by Claude Code. Handy to see which sub-agent is the expensive one. It is read from the session files, not an estimate of your bill.
• Stuck or quiet agents — in the live call, a tile pulses amber with ⏳ when a command has been running for over a minute, shows 💤 when a working agent has gone quiet for 3 minutes, and a red ✗ counter for failed tool calls. The browser tab title shows how many agents are working.
• Built-in demo — no session yet? ⚙ → Source → Demo plays a fictional team (no data of yours).
• Live call — follows your newest session (Claude Code, Codex or Mistral Vibe) in real time. Each agent is a small character at its desk; the one who speaks gets the spotlight and keeps it until someone else talks. Agents never talk over each other: when several have something to say at once, they take turns, so every line is heard.
• Real voices, lip-sync — lines are spoken aloud by a small speech model on your machine (a one-time install); mouths follow the actual volume of the voice. Until then, and if you prefer it, the agents babble (see Voices).
• Live screen sharing with the real output — when an agent works, its screen fills the stage (presenter's camera in the corner, "sharing" badge): a terminal with the command history and the actual output of each command, search results and tool calls; or an editor with the file it reads (real content and line numbers) or the exact diff it applies. See What you see for each agent below.
• Agents come and go — a sub-agent joins when it is given a task and leaves the call 2 minutes (playback time) after its report, with a dissolve animation. It comes back if it gets a new task.
• Sub-agent tree — the full hierarchy (you → the lead → sub-agents → their sub-agents) with status, tool counts and the current action.
• Click any agent — its prompt, tool usage and reports.
• Replay — pick any past session, scrub through it, see a timeline of who worked when, and get a recap (agents, tool calls, commands, files changed, busiest agents).
• Video export — ⚙ → Tools → 🎞 Export video records the call at 1280×720 with voices and subtitles, as an MP4 (H.264 + AAC) when the browser can record it, and as a .webm when it cannot. X, phones and most chat apps refuse .webm.
• Clip to share — ⚙ → Tools → 🎬 Clip to share (25 s) records about 25 seconds, saved as agent-standup-clip.mp4 (.webm when the browser cannot record MP4). In the middle of a replay it starts at the current position; otherwise one spoken line before the first error or loop of the session, and failing that at the moment the lead first hands out work. The line being said when time is up is finished rather than cut, so the length varies by a few seconds. The clip carries the site address (atoll92.github.io/agent-standup) in the top right corner and ends with a short end card; the full export has neither.
• Hide code & file names in videos — ⚙ → Tools, off by default, for the full export and the clip. Commands, their output, file contents, diffs and search results on the shared screen are drawn as grey bars of the same length; file names are removed from the window title and from the activity tags on tiles (the tool's name, e.g. Bash or Read, stays); the session title is replaced by "Agent Standup"; pictures (screenshots, images) become an empty frame. What the agents say is not masked: speech bubbles, subtitles and voices are kept as they are, so check the clip before posting if agents quote code or names aloud.
• In French, Spanish or Portuguese too — the call can be held in French, Spanish or Brazilian Portuguese: agent text is translated on the fly and read by voices of that language (see below).
• Office music — optional live web radio (FIP and its themed stations) that fades down smoothly while an agent talks and fades back up when the room goes quiet.
• Office ambience, optional (off by default) and fully synthesised: keyboards that type when agents run commands, clicks when they read, a coffee machine, a phone buzzing, a far-away door… (⚙ → Office ambience). Plus small reactions and jokes between agents ("episode mode").

Claude Code plugin

You can also open the call from inside Claude Code. The plugin adds two skills: /agent-standup:open starts the app and opens the call in your browser, and /agent-standup:status says in one line who is working and who needs you.

/plugin install agent-standup --marketplace Atoll92/agent-standup

On Claude Code before 2.1.275: /plugin marketplace add Atoll92/agent-standup, then /plugin install agent-standup@double-geste. The plugin runs the same npm package as the command line, pinned to an exact version.

Codex sessions too

Sessions of OpenAI’s Codex (the desktop app and the codex command) appear next to your Claude Code sessions: in the session list (project “Codex”), on the board, and in the live call when one of them is the session you last typed in. Nothing to set up: the app reads the files Codex already writes in ~/.codex/sessions (or $CODEX_HOME), and never writes there.

• What is shown: what you typed, what the agent says, each command with its real output and whether it failed, files it created or changed, web searches, MCP calls, the questions it asks you, its reasoning summaries as inner thoughts, token counts, and the end of each turn.
• Not yet: Codex sub-agents (a Codex session is the lead and you), and pictures. The short reviews Codex runs on its own before a command are not shown as sessions.
• In VS Code the call follows the Codex sessions started in the folder you have open, like it does for Claude Code.
• Turn it off or point it elsewhere: in VS Code, the setting agentStandup.codex; anywhere else AGENT_STANDUP_CODEX=off, or AGENT_STANDUP_CODEX=/path/to/codex-home (command line: --codex off or --codex DIR). With --projects DIR, Codex sessions are only read if you also pass --codex.

Mistral Vibe sessions too

Sessions of Mistral Vibe (Mistral's coding agent, the vibe command) appear next to your Claude Code sessions: in the session list (project “Mistral Vibe”), on the board, and in the live call when one of them is the session you last typed in. Nothing to set up: the app reads the folders Vibe already writes in ~/.vibe/logs/session (or $VIBE_HOME), and never writes there.

• What is shown: what you typed, what the agent says, each command and tool call with its real output and whether it failed, files it created or changed, its todo list, its reasoning as inner thoughts, token totals, the end of each turn, and its sub-agents: each one joins the call under the name the lead gave it, with its orders and its answers.
• Two engines, both read. Vibe 2.26 starts a new engine by default, which keeps the time of every event: the call is exact. With it a sub-agent leaves no record of its own, so you see what the lead asked it and what it answered, not each of its commands. The earlier engine (vibe --legacy-harness, and older versions) does record a sub-agent's own commands, but writes no times and saves at the end of each step: followed live the times are right to within a step, and a session opened afterwards is spread between its start and its end.
• On the board a Vibe session that waits for your approval or your answer shows as “needs you”, because Vibe says so itself.
• Not shown: pictures, and sessions kept somewhere else through session_logging.save_dir in Vibe's own settings (point AGENT_STANDUP_VIBE at a home folder that has logs/session in it).
• Turn it off or point it elsewhere: in VS Code, the setting agentStandup.vibe; anywhere else AGENT_STANDUP_VIBE=off, or AGENT_STANDUP_VIBE=/path/to/vibe-home (command line: --vibe off or --vibe DIR). With --projects DIR, Vibe sessions are only read if you also pass --vibe.

Checked against real sessions of Vibe 2.26.1, in both formats. Watch the 45-second video (en français).

Cowork sessions too

Sessions of Cowork (a tab of the Claude desktop app) appear next to your Claude Code sessions: in the session list (project “Cowork”), on the board, and in the live call when one of them is the most recent. Nothing to set up: Cowork records its sessions in the format of Claude Code, in the desktop app's own folder (local-agent-mode-sessions). Agent Standup reads them there and never writes there.

• What is shown: all that a Claude Code session shows, sub-agents included. A session has the name the desktop app gave it, its tab has a small “Cowork” mark, and in its call the lead reads “Lead · Cowork”.
• Scheduled tasks are sessions too. A task Cowork runs on a schedule appears when it runs. Nobody typed it, so it does not take the call from a session you are working in: the live call goes to it when nothing you typed in is still at work.
• Not the Chat tab. A conversation in the Chat tab of the desktop app is kept on Anthropic's servers, not on your computer: there is nothing for Agent Standup to read. The Code tab is Claude Code, and has always been shown.
• A window that follows one folder (VS Code open on a project) leaves Cowork sessions out of its call: a Cowork session is in no folder of yours. The Sessions view lists them everywhere.
• Where it looks: ~/Library/Application Support/Claude on macOS, %APPDATA%\Claude on Windows (and the app's own folder under %LOCALAPPDATA%\Packages when it is installed as a Windows package), ~/.config/Claude on Linux. Checked against real sessions on macOS; the Windows and Linux places are not tested yet.
• Turn it off or point it elsewhere: AGENT_STANDUP_COWORK=off, or AGENT_STANDUP_COWORK=/path/to/local-agent-mode-sessions (command line: --cowork off or --cowork DIR). With --projects DIR, Cowork sessions are only read if you also pass --cowork. There is no VS Code setting for it yet.

What you see for each agent

On its tile

• Activity tag (top right) — what the agent is doing right now: Reading App.jsx, Editing foo.ts, Writing …, Running npm test (or the command's own description), Searching “pattern”, Listing **/*.js, Fetching host, Updating todo list; then ✓ done when it reports.
• Ticker (bottom) — the current tool call and target, with a spinner and a running counter while it executes, then ✓ / ✗ and its real duration.
• Status border: orange while working, spotlight while speaking, the tile fades and dissolves 2 minutes after its report.

On the shared screen (appears when an agent uses a tool, stays while a command is still running, then ~4 s after the last action)

• Terminal — ❯ command (with its description), then its real output (first lines, errors in red, “… +N lines”), ● Grep(…) / ● Glob(…) with matches, a spinner for what is still running. Footer: ✓ Bash · 1.2 s or ⠋ Bash · running 3.4 s.
• Editor — for Read: the file with its real content and line numbers (read-only); for Edit: the removed/added lines; for Write: the new content, with the tool's result line.
• Pictures — when a tool gives the agent a picture (a screenshot it takes, an image file it reads: PNG, JPEG, GIF, WebP), the picture is shown on the shared screen, in a taller window, and stays up for several seconds (8 in a live call) whatever the agent does next. It is also in the agent's Recent activity and in exported videos. Pictures are read from the session file itself, never from your disk.
• Artifacts — when a Claude Code session publishes an artifact (a page on claude.ai), its author shares it in a browser window on the shared screen: the real page, shown whole. When the turn ends it is in the line of files (🌐); click it there, or click the window, to pin it: full size, usable, until you close it (✕ or Esc). A pinned page follows the next versions the session publishes, ‹ › step through the other artifacts of the session, and ↗ opens the page on claude.ai. The page is put together from the session file (the Write and Edit calls that made it), never from your disk and never from claude.ai, and runs in a frame that can reach neither the app nor the network. A page that uses libraries kept on a CDN shows Load its libraries (also ⚙ → Display → Artifacts: load their libraries), off by default. An artifact made from a claude.ai template (a Claude Design booklet, for instance) shows the page each publish sent; one of several pages is also shown whole, every page at once when the lead says it is done, and laid out as on the claude.ai canvas when you pin it. One that was sent no page (a deck or a document filled with data), or whose page a command wrote, has no preview: the window shows its title and its link.
• Task list — if the agent keeps a todo list, a side panel shows it live (☐ pending, ◐ in progress, ☑ done).
• The presenter's camera lip-syncs when they speak; the subtitle bar shows what they say.

Click an agent — its prompt, tool counts and a Recent activity list (last 30 calls with status, duration, and expandable command + output).

Outputs are trimmed (about 1–2 KB each) to keep the call fast. Results of internal calls (sub-agent launches, messages) are not shown.

Screenshots

Personality, without noise

In a live call, an agent that has been at work for a while thinks to itself in a lighter bubble that is not said out loud: its own reasoning when the session keeps it, otherwise what it is on (the command it waits for, the files it goes between, its to-do item); about one in five is mumbled, low. In a live call, when the lead starts on its tools without a word after your order, it says what it is starting with ("Got it. Let me look at server.js."). Sub-agents answer when they get an order ("Running the tests, as usual.", "Grepping like there's no tomorrow."), say goodbye when they leave, and now and then have a little thought; a failing tool call makes the tile shake; confetti when the lead finishes a long job. Whoever receives a report sometimes says thanks ("Got it, thanks."), and an agent that has been thinking for a long time says so ("Hmm, let me think."). When the call has been silent for a while, someone fills the gap the way people do in video calls ("Can you hear me?", "You guys still here?", "Can everyone see my screen?"): after 45 seconds in a live call, about 10 seconds in a replay, at most three times per silence and further and further apart. It is all local (no API), absent from the transcript, and can be turned off in ⚙ → Little lines & animations.

Sub-agents also talk to each other, from what really happened: a new agent asks a question and the lead answers, a report gets a reaction that fits (all good, errors along the way, it took long) and sometimes a word from a third agent, an agent that fails three times in a row gets a “you okay?”, and every few minutes the lead calls a quick round. ⚙ → Display → Chatter between agents: Quiet (only the real exchanges) / Normal (default) / Lively. When the lead finishes, it reads the part of its last message that the live reading had cut short, so you know what was done.

Glance without the panel (VS Code)

The status bar shows live counts for your workspace (needs you, working, your turn) and turns red when an agent may be waiting for you; hover for the sessions, click to open the call on the most urgent one (agentStandup.statusBar).

Context gauge, highlights, loop watch

The header shows how full the lead's context is (ctx 62%, orange then red, with a warning before auto-compaction); tiles and list rows show it from 50%. In Replay, the Highlights bar jumps to the first error, the longest wait, compactions and possible loops. A 🔁 badge marks an agent repeating the exact same call or failing repeatedly. The context window (200k or 1M) is inferred, because transcripts do not state it.

Works with your tools

Codex and Mistral Vibe sessions are read as well (see Codex sessions too and Mistral Vibe sessions too), and so are the Cowork sessions of the Claude desktop app (see Cowork sessions too). Other agent tools (Gemini CLI, OpenCode…) are researched in ROADMAP.md: their log formats are mostly undocumented, so no adapter is shipped for them yet.

Languages

Pick one in ⚙ → Language: English, Français, 繁體中文, 简体中文, Español, Deutsch, Dansk, Português (Brasil), 日本語, 한국어. The interface (menus, chips, tooltips, recap) is translated; VS Code's command titles, setting descriptions and status bar follow VS Code's own display language. English and French are written by hand; the other languages are AI-drafted and not yet reviewed by native speakers, so corrections are very welcome: each language is one small JSON file (public/locales/<code>.json), see public/locales/README.md to fix or add one (npm run i18n checks the files). The website and the landing page exist in the same ten languages, with the same caveat; their files are in website/locales/ and landing/locales/. The words the agents say stay in the language of your session, unless the call is held in French, Spanish or Brazilian Portuguese (see Translation); the little banter lines exist in these three languages and in English, the demo in English and French.

Your own characters

Your own agent names. A sub-agent you defined yourself (.claude/agents/code-reviewer.md) or named when it was launched keeps that name on its tile, in the transcript and in the voices: "Code Reviewer" rather than a first name drawn from the list. Two of the same type become "Code Reviewer" and "Code Reviewer 2". The agent types Claude Code ships (general-purpose, Explore, Plan…) still get a first name. ⚙ → Characters → Keep my agents’ own names turns it off.

A cat pack ships with the app: ⚙ → Characters → Cat pack (11 cats by a contributor) puts them on the call at once, and Drawn characters goes back to the default. My folder uses your own images below (picked by itself the first time if your folder already has images).

Replace the drawn characters with your own pixel art. Drop PNG (or GIF/WebP) files, 300 KB at most each, into ~/.agent-standup/characters (⚙ → Characters shows the folder, and in VS Code Open folder opens it; the agentStandup.charactersPath setting changes it):

• lead.png (or claude.png) is the lead, you.png is you, any other image becomes a sub-agent (one per agent: two sub-agents only share an image once every image is taken);
• name-talk.png next to name.png is an optional open-mouth frame that shows while the character speaks;
• 32×32 or 64×64 px, transparent background, facing front, feet at the bottom looks best. The art is shown crisp (no smoothing) on top of the usual scene and desk, and every status, hand and badge still works. Your characters sit at the desk without the laptop (it would cover the bottom of the sprite); ⚙ → Characters → Laptop in front of your own characters puts a darker one back while they work. Press Reload in the menu to see changes. Three example images to start from are in examples/characters (in the repository, not shipped with the app), and the cat pack is in examples/cats.

Agents list

On wide windows (900 px+), a fixed list sits left of the call: every agent with its role and one status line, most urgent first (needs you, stalled, working, waiting, done) and the lead pinned on top. It stays readable with dozens of agents, which a grid of tiles does not. Click a name to put that agent in the spotlight (kept for 25 s even when someone else speaks), double-click for details. Each line carries what the agent has used so far (⚡). An agent whose tile has left the stage keeps its line at the bottom, under “done”, with its count: click it for its details. Toggle it in ⚙ → Agents list.

All your sessions (Sessions view)

The Sessions tab lists every session active in the last 12 hours, across all projects, in sections by state, what waits for you first: needs you?, your turn, working, waiting on sub-agents, idle (a card also says how many sub-agents are active). A session that waits for you says for how long, apart from how long it worked before (“worked 25 min · waiting on you for 12 min”), and one at work how long it has been at it. It refreshes every few seconds and reads only the end of each transcript, so thirty parallel sessions stay cheap. Click a card to follow that session in the call.

• The question on the card. A session that stopped on a question shows the question itself, so you can tell what it wants without opening it.
• Done. ✓ on a card whose turn is over, or that is idle, puts it away in a Done section at the bottom, closed until you open it. It comes back by itself when the session moves again; ↩ brings it back at once. Its tab leaves the row of tabs of the call too, unless the call is in that session. This is kept in a small file beside your settings, so every window agrees within a few seconds. The indicators (menu bar, status bar) still count the session.
• Its pull request. When Claude Code noted a pull request for a session, the card has a PR #n button that opens it on GitHub. Only pull requests on github.com are offered.

Several sessions at once. When a second session is active, a row of tabs appears under the header of the live call: it names the session the call is in and the others, each with its state. With a single session there is no row. The call is in the session you last touched: start a session or type in one, and the call goes there. Click a tab to pin the call there (📌): it stays while you go through the Claude Code tabs of your editor and while you type in it, until you ask something in another session, which takes the call there, or until you click that tab again. What the agents write does not move the call, so two sessions at work do not pull it back and forth. One case moves it without you: when the session you last typed in has finished and been quiet for a minute and a half while another one is still at work, the call goes to that one (a session you went to by a click is not left that way). When another session stops for you (a question, an edit that waits for your go-ahead, or its turn is over), a visitor leans in at the edge of the call, says which one, and takes you there on a click (⚙ → Display to turn it off). In VS Code the call also goes to the Claude Code tab you switch to, so a conversation and its call can sit side by side; a tab is matched by its title, so a conversation that has no title yet, the sidebar chat and a terminal do not move the call. Switching tab counts as going there by hand: the call stays with that conversation until you type somewhere else.

Files made for you. When a turn ends, the files the session created for you show in one line under the call: those the lead names, else the documents (a report, a picture, a spreadsheet), not every source file it wrote. A document made by a command (an export, a report written by a script) shows too when the lead names it. Click one to see it in its folder; nothing is read and nothing is run. The whole list is behind “+N” (⚙ → Display to turn the line off).

At a glance

Every tile carries one status: working (with what it is doing), waiting on another agent (who), needs you? (a permission or a question is open in the session, or a file edit or write has been pending for 15 s+), with how long it has waited and, in the header, the question or the command it waits on, stalled (waiting on an agent that has been silent for 2 min+), or done. A summary sits in the header, and the tab title shows how many agents need you. Recent Claude Code notes in ~/.claude/sessions when a menu waits for your answer (a permission, a question), and the app reads it: a Bash command waiting for your go-ahead shows “needs you?” as soon as the menu opens, with the command (Bash(npm run build)), where the transcript alone makes it look like a command that runs. Where Claude Code does not note it, the signal is a heuristic: commands are not flagged because they can legitimately run for minutes (they get a ⏳ badge after 1 min instead), and edits that stay pending for 15 s+ are rare on real sessions, so it is mostly right.

Voices

With the real voices, each line is read with a voice of its own language (English, French, Spanish or Portuguese), whatever language the interface is in. A line in a language the model does not speak (German, Danish, Japanese, Korean, Chinese) is read by a system voice of that language when your computer has one.

Two choices in ⚙ → Voice style (default: Natural voices):

• Animal Crossing — every agent "talks" in gibberish, one blip per letter, with a sentence melody (up for questions, down for statements) and a register of its own.
• Natural voices — Kokoro-82M, a small open speech model that runs on your machine. The real words, and the default: click Install in ⚙ (about 750 MB, one folder in your user cache, installed with npm; the model is fetched on first use, and one download covers every language). Until it is ready the agents babble, Animal Crossing style, and a line at the bottom of the call offers the install in one click (close it and it does not come back). It speaks English, French, Spanish and Brazilian Portuguese with eight voices each. The model has few voices outside English (one French, three Spanish, three Portuguese): the lead takes one as it is, and the other characters get one of them mixed with their English voice, so each agent sounds like itself in every language. These three languages need a phonemizer the install adds (eSpeak NG, 18 MB, GPL-3.0, fetched by npm like the rest and not shipped with the app); if you installed the voices before 1.32.0, ⚙ → Voice style offers to add them when the interface is in one of these languages, and until then their lines are read by your system voices.

Our tip: install the natural voices once you are set up. The agents then sound like real people, which is the nicest way to follow a call.

Up to 1.31.0 there were five styles, and the natural voices were called Very natural voices. The Cartoon voices (your system voices with an effect, the default then), the plain system voices and Simlish are gone. If you had Simlish you are now on Animal Crossing; if you had the cartoon or the plain system voices you are on the natural voices, which babble until they are installed.

Long lines are subtitled one sentence at a time, in step with the voice.

Handy extras

• Copy as Markdown in the recap: a ready-to-paste standup report (who did what, files, tokens, final report of each agent).
• Notify me when agents finish (⚙): a soft chime, the tab title and, in VS Code, a notification once the team has worked for a while and gone quiet.
• Search the transcript, ambience volume, and a compact layout when the panel is narrow (side bar): under about 300 px wide, one row per agent instead of tiles. In replay, speed presets (1×, 2×, 4×, 8×) and each session remembers where you stopped.

Install

Four ways in — pick one (details in Works with your tools): the VS Code extension (below), the agent-standup command (npx @double_geste/agent-standup --open), the desktop app (macOS, from the release page), or the Claude Code plugin. Cursor, Windsurf and VSCodium get the same extension from Open VSX: search Agent Standup in their Extensions panel.

VS Code

Option 0 — from the Marketplace (easiest)

In VS Code: Extensions view → search Agent Standup → Install. Or: https://marketplace.visualstudio.com/items?itemName=DoubleGeste.agent-standup

Then click the Agent Standup icon in the Activity Bar (or run Agent Standup: Open live call) and press Join.

Option A — from a GitHub Release

1. Open the repository's Releases page and download agent-standup-<version>.vsix (don't send it by email: Gmail and others block .vsix files).
2. In VS Code: Extensions view → … menu (top right) → Install from VSIX… → pick the file. Or from a terminal: code --install-extension agent-standup-<version>.vsix (macOS: run Shell Command: Install 'code' command in PATH first if code isn't found).
3. Developer: Reload Window, then click the Agent Standup icon in the Activity Bar (or run Agent Standup: Open live call) and press Join.

Direct download of the latest version: https://github.com/Atoll92/agent-standup/releases/latest/download/agent-standup.vsix
Or with the GitHub CLI: gh release download --repo Atoll92/agent-standup --pattern "*.vsix", then the code --install-extension line above.

Option B — build it yourself

git clone https://github.com/Atoll92/agent-standup.git
cd agent-standup
npm run package                       # needs Node 18+; produces agent-standup-<version>.vsix
code --install-extension agent-standup-*.vsix

To hack on it instead: open the folder in VS Code and press F5 (launches an Extension Development Host).

Option C — no extension

After cloning the repository (Option B): node server.js and open http://localhost:4747 (see Standalone below). Without cloning: npx @double_geste/agent-standup --open.

The first click on Join is required by browsers to allow sound.

Commands

Settings

Controls

In Replay and in the demo, a slider and a speed control appear at the bottom, and clicking a transcript line jumps there.

Office music

Turn on Office music in ⚙ and pick a station (FIP, Groove, Jazz, Electro, Reggae, Rock, Nouveautés, Pop) and a volume. The stream is played live from Radio France's public Icecast servers directly in the page; it starts when you press Join (browsers require a click). When any agent speaks, the music drops to ~15 % within a fraction of a second and returns slowly after the last line. Muting (🔊 / M) stops the stream. Music is for your ears only: it is never included in video exports. Stations are for personal listening; nothing is stored or redistributed.

One window talks at a time

If several Agent Standup windows are open on the same machine (sidebar, editor panel, browser tab), only the last one where you pressed Join / Play / 🔊 plays sound; the others stay silent and show 🔈 (click it to take the sound back). This avoids hearing the same call twice, possibly in two languages. A window that stops answering releases the sound after a few seconds.

System voices (per OS)

The system voices are only a fallback for the natural voices: for a line in a language the model does not speak, and for a line it could not make. They are generated on your machine:

If no voice of that language is available the line goes to an English voice.

Translation (French, Spanish, Brazilian Portuguese)

Agent messages are in the language of the session. With French selected, each line is translated as it arrives using your local claude CLI (claude -p, Haiku) through two warm background workers: about 3–4 s per line once warm (the very first line takes ~10 s), then spoken. If a translation is not ready in time (7 s live), that line is read in the original language. The menu shows a hint if translation or the French voice is unavailable.

Spanish and Brazilian Portuguese work the same way: with one of them as the interface language, the call is held in it. The first time, a line at the bottom of the call says that it is translated with your Claude Code, with Turn off next to it, because translating sends the lines to Anthropic through your Claude Code login and counts on your plan. The switch stays in ⚙ → Language (Call in this language). Off, the call is the session as it is; a session that is already in Spanish or Portuguese is read by voices of that language either way.

Optional speed-up: set ANTHROPIC_API_KEY in the environment of VS Code / the server and translation calls the Anthropic API directly (~1–2 s).

Privacy & security: everything runs locally. The server listens on 127.0.0.1 only and rejects requests that do not come from localhost (other websites and DNS-rebinding attempts get a 403), so no web page can read your sessions. Translation and jokes send the shortened agent lines to Claude through your own Claude Code login (or your API key if you set one). With translation, jokes and music off, which is the default, nothing leaves your machine. (Office music connects your browser to icecast.radiofrance.fr while it is on.)

Standalone (no VS Code)

node server.js            # http://localhost:4747   (PORT=… to change)

• http://localhost:4747/ → live, newest session
• http://localhost:4747/?live=0 → replay picker
• ?proj=<project-dir> → limit live to one project folder under ~/.claude/projects

Requires Node 18+.

How it works

Claude Code writes every session to ~/.claude/projects/<project>/<session>.jsonl, and each sub-agent to <session>/subagents/agent-*.jsonl (with a .meta.json linking it to the tool call that created it). Agent Standup merges these into one timeline: orders (the prompt a lead sends to a sub-agent), spoken text, tool calls (with their inputs), the results of those calls (matched by tool-call id, with real durations) and final reports. Each file is parsed once and then only the new lines are read, so even a very large session refreshes in a few milliseconds; responses are compressed. The page polls the server every 0.7 s and plays new events.

The website at agent-standup.com counts its visits with Vercel Web Analytics, the counter of its host: no cookie, nothing loaded from another service, and no count for a browser that sends Do Not Track. The earlier landing page on GitHub Pages counts visits anonymously (a 1-pixel file hosted on GitHub per channel: no cookies, no third-party tracker) and asks GitHub which release is the latest. The app itself has no telemetry. Details in the privacy policy.

Troubleshooting

• Nothing happens / "Waiting for a session" — start Claude Code in the open folder; with onlyThisWorkspace the call follows that folder.
• Panel is blank — use Open in browser; if you set a fixed agentStandup.port it may be taken (use 0).
• Two voices at once (e.g. French and English) — update to ≥ 1.8.1: only one window plays sound now. A line that could not be translated in time is read by an English voice, not by a French voice reading English.
• No sound — press Join (or Play) once; check 🔊 isn't muted and Agent voices is on in ⚙.
• French is read in English / hint about translation — Agent Standup looks for Claude Code in your PATH, the usual install folders (Homebrew, npm, nvm, ~/.local/bin, Windows %APPDATA%\npm) and the copy bundled with the official VS Code Claude Code extension. If none is found the app quietly stays in English; set agentStandup.claudePath, or run claude once in a terminal to sign in. If VS Code itself was started from inside a Claude Code session, restart it normally.
• Characters or backgrounds look empty — update to ≥ 1.2.1 (fixed a duplicate SVG id bug).
• Terminal output missing for a command — the call is still running (spinner), or it is an internal call (sub-agent launch / message); outputs are only recorded once a call finishes.
• Pill says OFFLINE — the local server stopped answering (another VS Code window that was serving the port was closed: the extension takes over within a few seconds), or restart node server.js.
• Delay behind the real session — expected ≈ 5–7 s when the call is translated (French, Spanish, Portuguese), ≈ 2 s otherwise; the call skips ahead when it falls too far behind.
• Something else broke — ⚙ → Report a problem shows what the app logged on your computer and opens a pre-filled issue for you to send.

Limits

• Tool outputs are trimmed (about 1–2 KB each); the shared screen shows the beginning of long outputs.
• Output only exists once a tool call has finished: a running command shows a spinner and its live timer, not partial output.
• The natural voices need about 750 MB and npm; without them the agents babble.
• Very large sessions (10 MB+ of transcript) take a second or two to load the first time (about 1.9 s for a 12 MB, 168-sub-agent session on a recent Mac); after that updates take milliseconds.
• Video export and the clip record in real time and need a Chromium-based browser/VS Code. MP4 recording was checked in Chrome; it has not been tested inside VS Code's own panel, which may still produce a .webm.
• "Hide code & file names in videos" does not mask what the agents say (speech bubbles, subtitles, voices).

Credits

Several ideas in 1.23 (settings kept in a file, session stickiness, subtitles that follow the voice, Kokoro voices) started in neondragon's fork of this project. Thank you! Their code was read and rewritten for this codebase. rqwj made each sub-agent get its own character image (#1) and KajetanCiunelis drew the cat pack in examples/cats and suggested the laptop option (#2). The Traditional Chinese wording was polished with suggestions from a Taiwanese reader on Reddit. jainefer added the Korean interface (#5). Unofficial: not affiliated with Anthropic, OpenAI or Mistral AI.

License

MIT

Support the project / work with me

This dev was replaced by his own agents. If Agent Standup made your sessions more fun, you can buy him a coffee ☕ — there is also a ☕ line in the ⚙ menu and at the end of each recap.

Agent Standup is free and MIT licensed. I'm also available for work around AI agents and developer tooling: reach me through my GitHub profile, Atoll92.

Settings menu and shortcuts

⚙ has sections that fold (and remember it), a search box, and a Display section. “Notify me when the agents finish” is on by default (change it in ⚙ → Sound). The ? help also explains what each status means.

Press ? in the app for the full list. Highlights: Space pause, M mute, T tree view, J / K next agent that needs you, Cmd/Ctrl+K go to a session, and in replay ← / → step, N next highlight.

Name and logo

The code is MIT licensed: use it, change it, fork it. The name, the logo and the characters are how people recognise this project, and they are not there to brand something else. TRADEMARKS.md says what is fine (writing about it, building on it, shipping it unmodified) and what needs asking first (releasing a fork under this name, using the logo or the characters for another product). The artwork and its colours are in brand/.

Security and what it reads

The full privacy policy says what is read, what is kept on your computer and what can leave it.

• It reads local files and sends nothing by default. Agent Standup reads the session files Claude Code already writes (~/.claude/projects, or CLAUDE_CONFIG_DIR, and the status it keeps for each open session in ~/.claude/sessions) and the ones Codex, Mistral Vibe and Cowork write (~/.codex/sessions, or CODEX_HOME; ~/.vibe/logs/session, or VIBE_HOME; the local-agent-mode-sessions folder of the Claude desktop app; each can be turned off) and draws them in a page. It never writes to your sessions; the only time it starts Claude Code is the optional translation helper described below.
• Local only. The server listens on 127.0.0.1 (never on the network), refuses requests whose Host or Origin is not local, and refuses cross-site requests on every endpoint that does something (translation, jokes, sound, settings).
• What calls out: translation, which is on when the interface is in French, Spanish or Brazilian Portuguese (for the last two it can be switched off), and three things that are optional and off by default: jokes, the FIP radio stream, and the usage counter (one empty request a day to a 1-pixel file, only if you turn it on). Translation and jokes go through your own Claude login (the small Haiku model, with all tools disabled); when one of them is on, the lines of a Codex session are sent the same way, to Anthropic.
• No dependencies: the code is plain Node.js and one HTML file, so there is nothing hidden in node_modules. The optional Kokoro voices install a third-party package only when you click Install.
• Many sessions at once? The Sessions view shows every session across your projects on one board (working, waiting, needs you, done); click one to open its call. Live follows one session at a time, the most recent or the project you pick.
• Found a problem? ⚙ → Report a problem opens the draft of an issue on GitHub, filled with the app's version, your system and the last errors the app wrote to its local log (errors.log, beside your settings). You see the text first, and nothing is sent until you send the issue yourself.

Privacy and usage counting

Agent Standup reads local session files. With the interface in French, Spanish or Brazilian Portuguese it translates the call through your own Claude login (for the last two this can be switched off); in any other language it sends nothing anywhere by default. The other network uses are optional: jokes (your own Claude login, off unless you turn them on), the FIP radio stream (if you turn it on), and the optional usage counter: if you set agentStandup.shareUsage to true (off by default, and ignored when VS Code telemetry is off), the extension makes at most one empty request per day to a 1-pixel file in this repository's releases. There is no identifier and no content; GitHub sees your IP like for any download. It lets the maintainer count active installs (npm run stats).