All Features

Everything AgentsRoom does for you

Each feature is built to save you time and keep you in control. Browse them all below.

Video demos

See every feature in action

Each feature below ships with a short video walkthrough. Click the play button on any card to start watching.

01Orchestration

Run a fleet of agents

AgentsRoom multi-project multi-agent: visual cockpit with parallel Claude Code, Codex and Antigravity AI agents across multiple projects, color-coded roles, live status, per-project counters
Multi-project · Multi-agent

Many projects, many AI agents, one window

The visual cockpit to drive Claude, Codex, GitHub Copilot CLI, Cursor and 15 other agent CLIs in parallel across all your projects at the same time. Color-coded roles (DevOps orange, Frontend cyan, QA red, Architect blue), animated status dot on every agent card, working/done counters per project, pulsing red flag the moment an agent gets stuck. You know in one second which agent is working on which project, which one finished, which one needs your reply. No more 12 terminal tabs, no more guessing where you left off. The multi-project, multi-agent grid your Claude Code sessions always deserved.

Learn more
Idea Radar
Sorting
Raw signals

“cannot find my old exports”

support chat

“export is buried in a submenu”

public form

“please add dark mode”

demoted ticket
Clustering
Themesreporters
Exports14
Appearance9
Mobile5
1Capture the verbatim
2Cluster by theme
3Count the people
Idea Radar

The idea board that sorts your user feedback

Throw support messages, feature requests and half-formed notes into the Radar and they come back grouped by theme, deduplicated, and counted by the number of people who asked. Read the same corpus four ways: themes, impact/effort matrix, opportunity solution tree, semantic map. Users vote and comment on the public board, an AI product manager proposes the next ticket, and a ripe idea becomes a backlog ticket while a ticket you will not build goes back to being an idea.

Learn more
AgentsRoom backlog task board: kanban board to orchestrate AI agents with drag-and-drop from TODO to Development to Done
Task Board

Drag a task, spawn an agent

Kanban board for your AI agents. Create tasks, drag them to Development, and an agent starts working immediately. Give any column its own prompt (clarify, plan, review) and the board becomes a pipeline. Track every feature visually: who does what, what is done, what is next.

Learn more
AgentsRoom Remote Backlog: public feedback portal with client tickets, upvotes and a roadmap wired to Claude Code AI agents
Remote public backlog

Your clients drive the backlog, your agents ship it

Expose your project backlog as a public or private feedback portal. Clients submit bugs, request features, upvote tickets, leave comments, and every ticket lands directly in the same AgentsRoom backlog your AI agents execute against. Feedback-driven AI development, widget + Chrome extension included.

Learn more
Idea scoping
Scoping
Feedback

“Add a button somewhere on the profile screen”

Scope this idea
PM agent
Your product, mocked
New button
PM: Does this match what you had in mind?
1Read the codebase
2Build the mockup
3Publish for review
Ticket scoping

Scope fuzzy feedback into a mockup before you build

One click on a feedback ticket spawns a Product Manager AI agent that reads your codebase and builds a mockup faithful to your real product, with the requested change in context. Validate the idea and clarify it in a two-way conversation on the same ticket, all before any development starts. Scoping never starts the task and never sends a notification.

Learn more
AgentsRoom Teams: visual multi-agent workflow editor where Claude Code agents pass work between Dev, QA and PM roles like a real software team
Multi-Agent Workflow

Agent Teams: a real tech crew, scripted

Chain your AI agents like a real engineering team. Drop nodes on a visual canvas (Dev, QA, PM, Security, DevOps), connect them, and AgentsRoom orchestrates the handoff: feature summary, diff, risks, test hints. Pick how the roles run: one agent that morphs from role to role and keeps the whole context, or one live agent per role that talk to each other while they work, in any direction. Conditional edges, feedback loops, max-cycles guard. The n8n-style workflow editor your AI dev team was missing.

Learn more
Agent mail
1 unread
Backend Dev
Claude Code
Checkout flow ready for review
QA Engineer
CodexInbox
Stored
Queued
Delivered
Read

Recipient busy, message held

Agent messaging

Your agents write to each other, from any CLI

The saved agents of a project become permanent members with a real address and a real inbox. Any of them can write to another, whatever CLI it runs on, and the message is stored on disk before delivery is even attempted. An agent that is offline, a CLI that crashes or an app restart never loses one: it waits, and it arrives, with a receipt that says it was actually read.

Learn more
Agent Morphing
Same session
Morph to
Product Manager
Fullstack Developer
QA Engineer
Product Manager
Live agent
Context100%
TrailPMDevQA
Live role switching

Agent Morphing: switch roles, keep the context

Change your running agent's role on the fly. PM becomes Fullstack Developer becomes QA, in the same session, with the full conversation context preserved: no respawn, no re-explaining. Morph into any of the 14 built-in roles, The Agency experts or community agents, with a fresh eyes option for critical reviews and a morph trail of every role the agent has been. One agent, a whole team, whatever CLI agent you run.

Learn more
Agent ForkWorking
Backend Dev
  • read auth/session.ts
  • ruled out the cookie approach
  • trying the token refresh
Agent Fork

Fork an agent, keep the whole conversation

One click on a running AI agent creates a twin that already has its entire conversation: same role, same CLI, same model, same context. On Claude, Codex and Grok it is the CLI's own fork of the session; everywhere else the twin starts fresh with the parent's transcript handed to it, and the app says so on screen. Run both consoles in parallel on separate conversations, or put a question to a read-only copy of the agent without taking its terminal.

Learn more
AgentsRoom Hand over and restart: the confirmation dialog with the first message of the next conversation, Read your handover and continue
Hand over and restart

A fresh context for your agent, without losing the thread

One click on a running AI agent and it writes its handover, where it stands, what it decided and what is left to do, in a file of the project. Then its console starts a new conversation in the same tab, and the first message tells it to read that file and carry on. No /clear to type, no prompt to paste, on Claude Code, Codex and every agent CLI that has the AgentsRoom tools.

Learn more
Learning Loop

Your corrections

  • No, run the tests first
  • I already said: not the lockfile

Proposed rule

Run the test suite before every commit

Accept

Every agent follows it

Learning Loop

Agents that learn from your corrections

The corrections you keep typing to your agents become short rules you validate in one click. Accepted rules go into the project memory, and every agent follows them from then on, on Claude Code, Codex and every other CLI.

Learn more
Agent roster
Share via chat
Built-in roles14 roles
FrontendBackendQA
The AgencySpecialists
DataMLCloud
CommunityShared by users
Rust proa11yPrisma
Install
Orchestration · Custom agents

Build, share and install AI coding agents

AgentsRoom is not one assistant, it is a whole roster of agents. Start with 14 built-in roles, add specialist agents from The Agency, or build your own custom agent with its own system prompt. Then share an agent with a teammate in one drag through the chat, or publish it to the community so everyone levels up together. Every agent works with Claude, Codex, GitHub Copilot CLI, Cursor or any of the 15 other agent CLIs AgentsRoom supports.

Learn more
Agent Suggestions
Advisor
What do you want to do?
write a launch blo
e.g. audit the login flow
Agent Suggestions

Describe the task, get the right agent

Not sure which agent to use? Type what you want to do and AgentsRoom suggests the best-suited agent: from your room, the built-in roles or 260+ catalog experts, each pick with a one-line reason. One click adds it, opens its terminal and pre-fills your task.

Learn more
Scheduled Tasks
On schedule
Review pull requests
Every day at 08:00Code Reviewer
Launching agent
Scan crash logs
Every hour at :15DevOps Engineer
Update dependencies
Monday at 09:00Full-Stack
Runs while you sleepAuto run
Scheduled Tasks

Schedule your AI agents like a cron, without the cron

Scheduled Tasks runs your AI coding agents on a recurring schedule, with no cron expression to write. Program a Claude Code or Codex session to review pull requests every morning at 8 and again at 6, scan crash logs every night, check dependencies every Monday or ship a weekly changelog. Pick a frequency (every N minutes, hourly, daily at one or several times of day, weekly, monthly, or on your AI plan's quota window), a prompt and an agent role. Each run spawns a background agent in your project, sends you a notification, catches up missed runs on the next launch and keeps a reopenable history. A cron alternative built for recurring AI development workflows, provider-agnostic across Claude Code, Codex and Antigravity.

Learn more
Webhook triggerListening
POSTGitHubpull_requestAdd rate limiting to the API
Signature verified
Filter matched
Code Reviewer
{{event.title}} = Add rate limiting to the API
Nothing runs while nothing happensStarts on event
Webhook Triggers

Your agent starts when something actually happens

Webhook Triggers start an AI coding agent on an outside event instead of a clock. Each trigger gets a public URL and a signing secret you paste into GitHub, GitLab, Slack, Linear, Sentry or any service that can POST JSON. A pull request opens, a build breaks, a crash is reported, and the agent is already reading it. The signature is verified, an optional filter decides what deserves a run, a burst limit keeps a noisy service from starting twenty agents, and the payload becomes prompt variables. An event that lands while the app is closed is queued and replayed on the next launch, and with several machines only one takes it. Event-driven automation for developers, without a polling loop burning tokens on the answer no.

Learn more
Interview

How will we know it is finished?

  • The page is online
  • The client approved it
  • Other
Mission page
  • Goal
  • Done when
  • Constraints
  • Decisions taken
  • Open questions
Tasks
  • Write the text
  • Find three references
  • Prepare the offer
LaunchChangeCancel
Orchestration · Mission interview

An AI assistant that interviews you, then splits the work

Press New mission and the project's assistant asks one question at a time, until it knows what done means. A page is written live next to the conversation, then tasks with the right assistant and model start one after the other after your click.

Learn more
Agent delegation in AgentsRoom : a dev agent (Codex) delegates the test to a child QA agent on a cheaper model, dev to QA handoff in the agent list
Agent delegation

Your dev agent delegates the test to a cheaper QA agent

When a feature is done, the dev agent (Claude Opus, Codex) calls a single MCP tool and hands the validation off to an ephemeral QA agent running on a cheaper model (Claude Haiku, Codex mini). The QA agent drives the embedded browser, asserts the result and reports a clean pass or fail. You stop paying premium prices for browser clicks, and you stop bloating the dev agent's context with screenshots and DOM dumps. The right model at the right price for each step.

Learn more
Started in a terminalWorking in a terminal
Terminal
> Refactor the billing module> Move the webhooks to a queue> Add the retry tests
AgentsRoom
Same conversation, whole history3 / 3
Orchestration · Terminal sessions

Resume the conversations you started in a terminal

Already using Claude Code, Codex or Antigravity in a terminal? Open that folder in AgentsRoom: under the agents, Started in a terminal lists the folder's conversations of the last 30 days, with what you asked and when. Click Resume and an agent continues the conversation where you left it, with the CLI's own resume command and its whole history. Also from the mobile app.

Learn more
One repository, one checkout per agent
Main checkoutmain
  • Backend agent.worktrees/fix-login-redirectPrepared
  • Frontend agent.worktrees/settings-dark-modeWorking
  • QA agent.worktrees/agents/qaReview
Orchestration · Git worktrees

A git worktree for every agent, prepared and reviewed

Two agents in the same checkout overwrite each other. Tick one box on a ticket, or pick a worktree when you add an agent, and it works in its own checkout on its own branch, under .worktrees/ in the project. Before it starts, AgentsRoom copies your .env files and installs the dependencies detected from the lockfile, so the first build works. When it is done, the Worktree selector of the Changes tab shows its diff, commits and pushes it, and marking the ticket Done can merge its branch back. It works the same for Claude, Codex, GitHub Copilot CLI, Cursor and 15 other agent CLIs.

Learn more
Compare on several agents

Fix the login redirect loop

main, your last commit

Claude Coderace-1-claude
  • 3 files
  • +42 -11
  • 6 min
Kept
Codexrace-2-codex
  • 5 files
  • +88 -23
  • 9 min
Keep this one
Antigravityrace-3-antigravity
  • 2 files
  • +31 -19
  • 4 min
Keep this one
Orchestration · Compare agents

Claude Code vs Codex on your own code

Give the same task to 2 to 4 CLIs or models in one gesture, each in its own git worktree cut from your HEAD. Your main checkout does not move. A card shows the files, lines and working time of each agent with its whole diff, and Keep this one merges the branch you choose. AgentsRoom does not score the results: you read the code and you decide.

Learn more
One project, three repositories
  • /work/web-appProject folder · git
  • /work/apiLinked folder · git
  • /work/shared-libLinked folder · git
At launch
Claude Code--add-dir=/work/api --add-dir=/work/shared-lib
Codex--add-dir=/work/api --add-dir=/work/shared-lib
GitHub Copilot CLI--add-dir=/work/api --add-dir=/work/shared-lib
OpenCodefolders listed in its brief
Orchestration · Linked folders

Give your agents several folders and repositories

Your app and its API in two repositories, a game and its port to a newer engine in two folders: link the second one to the project instead of moving it. Every agent you start in the project gets the linked folders, through --add-dir on Claude Code, Codex, GitHub Copilot CLI, Antigravity CLI, Kimi Code and Mistral Vibe, and in its brief for the other CLIs. Each folder keeps its own version control, and the repository picker of the Changes tab reviews, commits and pushes each one.

Learn more
New workspace
NamePlan the next trip
MissionFind three itineraries under 2,000 euros and keep the bookings in one file
AgentTravel planner (your agent)
Created by AgentsRoom

~/AgentsRoom/Personal/plan-the-next-trip

  • Mission in every agent brief
  • Agent memory and pinned skills
  • Backlog, teams, dev commands
  • No git repository needed
Orchestration · Non-coding workspaces

Give agents work that is not code

A support incident, a trip, a product research: create a workspace from a name, a mission and an agent, with no repository to point at. AgentsRoom creates the folder, the mission rides in the brief of every agent on every CLI, and your saved agents arrive with their memory and their pinned skills.

Learn more
Non-dev mode
Who are you?
I'm a developer
I'm not a developer

My assistants

  • Lead assistant
  • Writer
  • Marketing Specialist
  • Sales Representative
  • Researcher
  • Finance & Admin
  • Customer Support
  • Brainstormer
TasksPromptsTriggersMemory
Non-dev mode

AI assistants for people who don't code

Answer I'm not a developer at the first question and AgentsRoom drops projects and code tools for one room of AI assistants: a Lead assistant, seven assistants from Writer to Brainstormer and 17 business roles to add, with the prompt library, triggers, tasks and memory. They run on the agent CLI you already sign in to, and one chip at the bottom right switches back to dev mode.

Learn more
From Linear to your agents
Linear workspaces
Acme
Globex
Assigned to me
Backlog · TODO
ENG-142Claude Code
ENG-147Codex
WEB-31GitHub Copilot CLI
Read only: nothing written back
Orchestration · Linear import

Work your Linear issues with AI coding agents

Connect one or several Linear workspaces with a personal API key each, keep Assigned to me or filter by team, project and status, preview and import. Every issue becomes a backlog ticket with its description, its context and its attached files, ready for Claude Code, Codex, GitHub Copilot CLI or another agent CLI. The import only reads: nothing is ever written back to Linear.

Learn more
From your tracker to your agents
Sources
Jira
Linear
GitHub Issues
GitLab Issues
Notion
Azure DevOps
Preview, then import
Backlog · TODO
ACME-412Claude Code
ENG-142Codex
#88Cursor
Read only: nothing written back
Orchestration · Ticket import

Import your tickets into AI coding agents

Connect Jira Cloud, Linear, GitHub Issues or Notion, choose what to bring in, preview and import. Every ticket lands on the backlog with its description and its attachments, ready for Claude Code, Codex, GitHub Copilot CLI or another agent CLI, and a refresh keeps it up to date without duplicates. The import only reads: nothing is ever written back to the source tool.

Learn more
From Jira to your agents
Jira projects
ACME
MOB
assignee = currentUser()
Backlog · TODO
ACME-412Claude Code
ACME-415Codex
MOB-88Cursor
Read only: nothing written back
Orchestration · Jira import

Work your Jira issues with AI coding agents

Connect your Jira Cloud site with an API token, pick a project or write a JQL query, preview and import. Every issue becomes a backlog ticket with its description converted to Markdown and its attached files, ready for Claude Code, Codex, GitHub Copilot CLI or another agent CLI. The import only reads: nothing is ever written back to Jira.

Learn more
From GitHub Issues to your agents
Repositories
acme/web
acme/api
Open · label: bug
Backlog · TODO
#412Claude Code
#415Codex
#88Cursor
Read only: nothing written back
Orchestration · GitHub Issues import

Work your GitHub issues with AI coding agents

Connect GitHub through the GitHub CLI or a token, pick a repository, a state and labels, preview and import. Every issue becomes a backlog ticket with its Markdown body and its labels, and the agent runs on your own machine in the CLI you choose. The import only reads: nothing is ever written back to GitHub.

Learn more
From Azure DevOps to your agents
Azure DevOps projects
Fabrikam
Contoso
[System.AssignedTo] = @Me
Backlog · TODO
#4120Claude Code
#4127Codex
#882Cursor
Read only: nothing written back
Orchestration · Azure DevOps import

Work your Azure DevOps work items with AI coding agents

Connect your Azure DevOps organization with a personal access token, pick a project or write a WIQL query, preview and import. Every work item becomes a backlog ticket with its description, repro steps and acceptance criteria converted to Markdown and its attached files, ready for Claude Code, Codex, GitHub Copilot CLI or another agent CLI. The import only reads: nothing is ever written back to Azure DevOps.

Learn more
From YouTrack to your agents
YouTrack projects
WEB
APP
#Unresolved for: me
Backlog · TODO
WEB-412Claude Code
WEB-415Codex
APP-88Cursor
Read only: nothing written back
Orchestration · YouTrack import

Work your YouTrack issues with AI coding agents

Connect YouTrack Cloud or your self-hosted instance with a permanent token, pick a project or type any YouTrack search, preview and import. Every issue becomes a backlog ticket with its description, its filled custom fields and its attached files, ready for Claude Code, Codex, GitHub Copilot CLI or another agent CLI. The import only reads: nothing is ever written back to YouTrack.

Learn more
AgentsRoom Import tickets window with GitLab Issues next to GitHub Issues, JIRA, Notion, Linear and Azure DevOps: issues from one project on gitlab.com or your own instance
Orchestration · GitLab Issues import

Work your GitLab issues with AI coding agents

Connect gitlab.com or your self-hosted instance with the GitLab CLI or a personal access token, pick a project, a state and labels, preview and import. Every issue becomes a backlog ticket with its Markdown description, its labels and its attached files, ready for Claude Code, Codex, GitHub Copilot CLI or another agent CLI. The import only reads: nothing is ever written back to GitLab.

Learn more
02Communication

Talk to your agents your way

Ask on the sideThe agent sees none of this
Ask
A copy of this agent answers you
why did you drop the cooki
Ask on the side

Ask a busy agent without interrupting it

Click the round Ask on the side button of an agent's console and type a question: a read-only copy of its conversation answers in a bubble, in streaming, while the agent keeps working and its history stays untouched. Up to five tabs per agent, a badge when the bubble is closed and a faster model by default, on Claude Code, Codex, Grok, GitHub Copilot CLI, OpenCode, Mistral Vibe, Antigravity CLI and Kimi Code, from the desktop or the phone.

Learn more
Settings > Language > Agent replies

Behebe den Login-Fehler

Claude CodeFixed: the session cookie expired before the redirect.

¿Por qué falla el test?

CodexThe test waits for a mock that is never reset.

このPRをレビューして

Antigravity CLITwo comments on the PR: one blocker, one nit.
Agent reply language

Every agent answers in your language

Pick a reply language once in Settings > Language > Agent replies and every agent AgentsRoom launches answers in it, whatever language you write in: Claude Code, Codex, GitHub Copilot CLI, Cursor and 15 other agent CLIs, with notifications and tab titles in the same language. A message asking for another language, or a project file that pins one, still wins.

Learn more
AgentsRoom Sketch: a drawing canvas inside the agent composer to sketch mockups and annotate screenshots, then send the drawing to an AI coding agent
Sketch

Draw it, and the agent sees what you mean

Sketch is a drawing canvas built into the agent composer. Open a blank canvas to sketch a mockup or a diagram, or open a screenshot to annotate it: circle the bug, draw an arrow, highlight the region. The drawing is attached to your message and sent to your AI coding agent. A picture is worth a thousand words, so stop describing the visual bug and draw it, without ever leaving the terminal.

Learn more
AgentsRoom Screenshot to Agent: capture a screen region with a keyboard shortcut and the screenshot is auto-attached to the AI agent composer
Screenshot

Screenshot to agent, in one shortcut

Send a screenshot to your AI coding agent without ever leaving the keyboard. Hit the configurable shortcut (Cmd+Option+Shift+P by default, Ctrl+Alt+Shift+P on Windows and Linux) or the composer's screenshot button, drag a box around the bug, and the screen capture attaches itself to your message. No external screenshot app, no saving the file to disk, no dragging it into the composer. Capture the screen region, the agent sees it, you keep your flow.

Learn more
AgentsRoom voice dictation: a microphone button in the agent composer turns speech into a prompt, speech-to-text transcription with GPT-4o Transcribe and Whisper, live voice waveform, 21 languages
Voice · Dictation

Dictate your prompts, your agent codes

Voice dictation is built right into the agent composer. Click the microphone, speak your prompt, and the speech-to-text transcription drops into the draft at your cursor. No separate dictation app, no copy-paste between windows. There is no transcription model to pick, it is chosen from the length of your recording, dictate in any of 21 languages plus auto-detect, and watch a live voice waveform while you record. On mobile, speech recognition runs on-device and the text is relayed to your focused desktop agent. Typing a long, precise prompt takes minutes, dictating it takes seconds: more context to your agent, fewer clarification round-trips, fewer wasted tokens. Nothing is sent automatically, the transcript lands in the draft so you review, edit and send with Enter. Voice dictation for Claude, Codex, GitHub Copilot CLI, Cursor and 15 other agent CLIs, the fastest way to write a precise prompt.

Learn more
AgentsRoom Voice Mode: a two-way voice conversation with an AI coding agent, listening state, hands-free toggle, selectable reply voice, talk to your agent and hear its spoken replies
Voice · Mode

Talk to your agents, hear them talk back

Voice Mode is a two-way voice conversation with your AI coding agent. Tap once, speak, and the agent answers out loud in a natural voice, no typing, no reading walls of terminal output. Turn on hands-free and it keeps listening for your next turn, so you can pace the room, sip your coffee, or watch the build while you talk through the plan. You ask where the refactor stands, it tells you. You say run the tests and report back, it does and speaks the result. Pick the reply voice (alloy and more), keep a beep cue between turns, and let language auto-detect so you speak in your own words. Voice Mode runs on the AgentsRoom backend (speech-to-text in, text-to-speech out) and works the same with Claude, Codex, GitHub Copilot CLI, Cursor and 15 other agent CLIs. It is the most natural way to stay in the loop with a long-running agent without being chained to the keyboard.

Learn more
CHANGELOG.md
ListenSummary

Voice credits are now shared between dictation, Voice Mode and Read Aloud.

The terminal selection panel gains a Listen button for anything you highlight.

The status bar keeps a control chip for as long as a read is playing.

3:20
Read Aloud

Listen to your docs instead of reading them

Read Aloud is text to speech built into AgentsRoom. Click the speaker in the markdown preview and a README, a spec or a three-hundred-line changelog is read out loud in a natural synthetic voice. Highlight anything in a terminal and hit Listen. Right-click to hear your agent's last answer, pulled from its conversation transcript rather than scraped off the screen. Code blocks, markdown tables and bare file paths are stripped before a single word is spoken, so you hear prose and not punctuation. In a hurry? The same button offers a condensed read: an AI summary is written first and that summary is what you hear, sized to a spoken length you pick, ten seconds, thirty or a minute. A forty-minute changelog answered in ten seconds. Progress bar, time left, pause, resume and a control chip in the status bar that outlives the window you started from. It came from a dyslexic developer who was copy-pasting our markdown preview into an external text-to-speech tool, and it earns its keep on tired eyes too.

Learn more
Message queue
Queued
1Fix the login redirect loop2
2Add a dark mode toggle
3Write tests for the API client1
Type the next idea...QueueSend
Message Queue

Stack your messages, send them in order

A message queue built into the composer. Stack several prompts above the input and send them to your AI agent one at a time, in the order you choose, instead of one overloaded mega-prompt. Each queued message gets a short AI-generated title and keeps its screenshots and attachments. Drag to reorder, reopen to edit, send on demand. A queued message can also carry a clock, or follow your subscription instead of the wall clock: send it the moment the quota comes back, or shortly before a 5-hour, daily, weekly or monthly window resets, so a long code review runs on the credits that were about to expire.

Learn more
AgentsRoom question card above an AI agent's input box: a multiple-choice clarifying question with clickable options and a mockup preview
Structured questions

Your agent asks, you click, it waits

When an AI agent needs a decision, AgentsRoom shows its question in a card above the agent's input box: clickable options with a line on what each one implies, several picks when it makes sense, mockups to compare, your own words in the card's field. The agent waits for your answer instead of guessing, and you answer from the desktop, the mobile app or the lock screen. Works with Codex, Claude Code, Cursor, GitHub Copilot CLI and every agent CLI with MCP.

Learn more
AgentsRoom Mobile Companion: dictate prompts, attach files, and use the iOS and Android share extension to send screenshots to your desktop AI coding agent
Mobile companion

Your phone, a remote for your AI agents

The Mobile Companion turns your phone into a remote for your desktop AI agents. Dictate a prompt or attach a file to the focused composer over the end-to-end encrypted relay. Share screenshots from any iOS or Android app and choose their project, with the current project selected, then send to an agent or save in its backlog. A closed agent starts automatically. Backlog photos are stored with the task in your account, even when the desktop is off. Works with Claude Code, Codex CLI and other supported providers.

Learn more
Personal assistantagentsroom://bot
Listening…
Phone
encrypted relay
Assistant, ready
Desktop
Personal assistant

Your personal assistant, one sentence away from your phone

The assistant is the one agent of your AgentsRoom account you can reach from anywhere. Say "Talk to Bot" to Siri, press the iPhone Action button, tap an Android widget or tile, the Assistant tab of the app, or share a text to it: the app opens the assistant's screen with the microphone already listening, transcribes what you said with the AgentsRoom speech-to-text into the composer in case a name came out wrong, and sends it over the end-to-end encrypted relay to your desktop. There, the assistant runs through the agent CLI you already use, with your MCP connectors (mail, calendar, messaging) and the AgentsRoom tools that reach every project: it reads and files backlog tickets anywhere, launches an agent on one, drafts a reply in your voice, and confirms before anything visible to someone else. It starts a fresh conversation every day and, if you allow it, remembers you in a private memory of its own that your project agents never see. The answer comes back as a push, and when a long task ends the desktop calls you by your first name. Called "Personal assistant" by default, renamable, custom avatar, one settings page on both screens.

Learn more
AgentsRoom prompt library: save, organize and reuse Claude prompts per project
Productivity

Save and reuse your best prompts

Organize prompts in folders, per project. Share them with your team via git. Keep personal prompts private with automatic gitignore. Send any prompt to an agent in one click.

Learn more
AgentsRoom skills library: SKILL.md skills attached to tasks and agents, exportable to Cursor and Codex
New

Modular procedures, any runtime

Save Claude SKILL.md skills, attach them to tasks or agents, and export to Cursor, Codex, Windsurf or Aider on demand. Trigger-based routing means the right procedure loads automatically.

Learn more
Project Memory
Shared with the project
Agent session
Agent, Monday
Context window100%
Working
Project memory+1
architecture
decisions
pitfalls
conventions
features
Survives every session. Read by every agent.
Project memory

The shared brain of your codebase

Your agents write down what they learn: decisions, architecture, pitfalls, conventions. It survives the session, and every agent and teammate on the project reads the same memory.

Learn more
One folder, one memory, every agent
Obsidian vault folder
architecture.md
auth-decisions.md
release-runbook.md
Project Memory
architecture.md
auth-decisions.md
release-runbook.md
Agents
Claude Code
Codex
GitHub Copilot CLI
Conflict: your choice
Dev toolkit · Obsidian Vault Sync

Sync an Obsidian vault folder with your agents' memory

Pick one folder of your Obsidian vault and link it to a project. Notes edited in Obsidian become the memory Claude Code, Codex, GitHub Copilot CLI and every other agent CLI read before they start, and the notes the agents save return to the vault as Markdown with their [[links]]. Both ways or vault to memory only, no Obsidian plugin, conflicts kept for you to settle.

Learn more
AgentsRoom Scratchpad: the prompt editor of the agent composer, for drafting long prompts and sending them to that agent
Prompt Editor

Write your prompt like you write an email

A prompt editor pinned to the AgentsRoom footer. Draft long, structured prompts for your AI agents in a real text area. Your draft is always there when you come back, even after you quit the app. One click sends the whole thing to the focused agent, and one click saves a good draft as a reusable prompt for your team.

Learn more
One click, in the composer
What you type

fix the login redirect, it loops

What the agent receives

Goal Stop the redirect loop on sign-in and land the user on the page they asked for.

Context Next.js app router, auth in middleware.ts, the loop appears after a session expires.

Constraints Do not change the session duration. Keep the existing redirect parameter.

Done when A signed-out visit to a private page signs in and comes back to that page, once.

Prompt enhancer

One click turns your rough request into a real prompt

Type or dictate six words, press Enhance, and the box fills with a prompt that has a goal, the context, the constraints and a definition of done. It uses the project's name, the role of the agent you are writing to and the project memory notes that match your draft, replaces your text in place with an Undo, and sends nothing to the agent until you do. Also in backlog tickets, in the Prompt Library and in the mobile app.

Learn more
AgentsRoom linked prompts: a task prompt inheriting shared context from other prompts in the Prompt Library
Communication · Prompt chaining

Prompts that call other prompts

A prompt in your library can link to other prompts and pull their content in when it runs. Write the shared rules once (your stack, your conventions, your review checklist), then reference them from every task prompt instead of copying them again and drifting. Chains resolve at send time, so the agent receives one complete brief, and updating the shared block updates every prompt that inherits it.

Learn more
AgentsRoom Figma to AI agents: a captured Figma frame attached to a backlog ticket for a coding agent to implement
Communication · Design handoff

From a Figma frame to shipped UI

Capture any frame straight from the Figma canvas with the AgentsRoom Chrome extension, drop it on a backlog ticket or an agent prompt, and let the agent build it against your real components. The design travels as an image plus your written intent, so nothing depends on a Figma MCP server being up, authenticated or compatible with your file. The capture stays attached to the ticket, so the agent that picks it up later still sees exactly what was asked.

Learn more
03Monitoring

Stay in control, in real time

AgentsRoom split view: multiple AI coding agents side by side in split panes, drag-and-drop horizontal and vertical split screen with Claude Code, Codex and Antigravity sessions, color-coded per-pane status
Split view · Multi-pane

See every agent at once, side by side

Split your workspace into multiple panes and watch several AI coding agents work at the same time, side by side. Drag an agent from the agent list onto the terminal on screen and drop it on an edge: left or right splits into columns, top or bottom stacks into rows, iTerm2 style. Switch the whole layout between horizontal and vertical in one click. Each pane keeps its agent's role color, avatar and live status, the focused pane gets a highlight ring, and your split layout is restored exactly when you reopen the app. Run Claude Code on the backend in one pane, Codex on the frontend in another, an Antigravity QA agent in a third, and keep an eye on all of them without ever tab-hopping. The multi-pane split screen your parallel agents were missing.

Learn more
Same session, two faces
Conversation

Why did the checkout total ignore the coupon?

Backend
3 tool calls

The discount was applied after tax. I moved it before the tax step and added a test.

Conversation view

Your agent's terminal, read as a conversation

Click the floating Conversation button in an agent pane and the session becomes a chat, read from the transcript the CLI writes on disk: your messages on the right, the agent's answers signed with its name, tool calls folded, answers appearing as they are written. Claude Code, Codex, Grok Build, Mistral Vibe, oh-my-pi, Pi and Antigravity CLI, on the desktop and the phone. The CLI keeps running underneath.

Learn more
AgentsRoom agent status: live indicators showing which Claude Code agents are working, done, or waiting
Live status

See who is working, right now

Color-coded status indicators on every agent card. Yellow pulsing dot when thinking, green when done, red when waiting for your input. Elapsed time and current activity displayed in real time.

Learn more
AgentsRoom Dynamic Island: a floating always-on-top AI agent status indicator with a click-to-focus finish notification for Claude Code, Codex and Antigravity agents
Always on top

Your agents' status, always in sight

A floating, always-on-top island shows whether your Claude Code, Codex and Antigravity agents are idle, working or waiting for input, even when AgentsRoom is in the background. When an agent finishes, a notification surges next to the island. Click it and you land straight on the right agent in the right project. Stop alt-tabbing to check.

Learn more
AgentsRoom context drift canary warning: an AI coding agent is losing context after several turns, with a one-click restart and a compact-conversation tip
Context canary

Catch an AI agent before it starts hallucinating

AgentsRoom watches a built-in canary on every agent: the one-line status each agent writes at the end of every turn. When an agent stops updating it for two turns in a row, that is the classic early sign of context rot and degradation, so AgentsRoom raises a warning above the terminal before the hallucinations begin. One click restarts the agent on a clean context, and a tip reminds you to compact the conversation. Works with Claude Code, Codex and every provider.

Learn more
Process Guard
Scan: 1 / min
Full-Stack Dev, child processes
ripgrep18 MB
71%
tsc --noEmit1.4 GB
96%
search6.4 GB
3%
node2.1 GB
0%
2 stuck processes
6.4 GB held, 0% CPU
Stuck for 23 min, no progress
Started by Full-Stack Dev
End the process
Old enough gets measured. Big enough, or too many at once, gets flagged.Memory back instantly
Process Guard

The runaway process finder for AI coding agents

An AI agent spawns a real system process for every tool call it makes: a search, a build, a script. Most finish in seconds, some never do, and those keep holding gigabytes of RAM at 0% CPU until the machine crawls. On one measured session, seven stuck processes claimed 41.8 GB on a 16 GB laptop, swap sat at 95%, and ending them returned 15.5 GB instantly. Process Guard sweeps the child processes of your agents once a minute and flags one only when it is large, old AND idle on the processor, so a real build is never mistaken for a stuck one. It measures the true memory footprint, compressed and swapped-out pages included, because a stuck process can show 20 MB while holding 8 GB. Orphaned processes that outlived their agent are caught too. Your dev command terminals, the shell and the agent CLI itself are never touched, and nothing is terminated unless you ask.

Learn more
System sleep
  • Agentrefactoring the auth module
  • Dev commandnpm run build
  • Databaseexport, 2.1 GB of 3.4 GB
  • Manual holdkept awake for 4 h
Screen off, machine working
Sleep held
Keep awake

Keep your computer awake while AI agents work

Lock your computer during a long agent task and it goes to sleep a minute later, freezing the agent, the build and the database export with it. AgentsRoom holds system sleep while an agent works, a dev command's job is running or a database transfer is in flight, and lets go the moment nothing runs. The screen still turns off. A cup in the status bar shows when the machine is held and why, and one click keeps it awake on demand for 1, 4 or 8 hours or until you turn it off, like Caffeine or PowerToys Awake. Agents can ask for it over MCP too.

Learn more
Pausing project
  • FrontendFinishing its turn
  • BackendWriting its resume summary
  • QAPaused
Every agent is paused: you can close AgentsRoom
Project pause

Pause your AI coding agents and resume them later

Press Pause on a project and every agent stops once its current turn is over, never in the middle of an edit. A quick pause resumes the same conversation; a long pause asks each agent for a resume summary and restarts from it in a new conversation, which keeps the restart light on quota. A panel follows each agent and tells you when you can close AgentsRoom, and until you resume, no launch, message, scheduled task or trigger wakes the project up. You can also pause a single agent while the rest of the project keeps working, by hand or automatically after a chosen idle delay; it wakes when another agent, a scheduled task or you need it. Also from the phone and through your personal assistant.

Learn more
Adaptive Mode
Analyzing task
Translate a UI string
Best fit
Haiku
Sonnet
Opus
Cost per taskMore tasks per day
Smart model routing

The right AI model at every message, without restarting the agent

Adaptive Mode reads each message before it leaves and picks the most cost-effective model for it. A quick fix or a translation goes to a light model like Claude Haiku; a deep refactor or an architecture design goes to a heavyweight like Claude Opus. In Automatic mode, Claude Code and Codex agents change model with no restart and nothing to type; with the other CLIs, one click applies the suggestion. Works with Claude, Codex and Antigravity, on by default on new installs, nothing to install. Stop overpaying with a flagship model on trivial tasks: spend less per task and run more tasks per day on the same plan.

Learn more
Default reasoning effort
Claude CodeVery highclaude --effort xhigh
CodexHigh-c model_reasoning_effort=high
Kimi CodeLowKIMI_MODEL_THINKING_EFFORT=low
Who decides the level
  1. Agent or ticket
  2. Project
  3. Settings
  4. CLI default
Reasoning effort

A default reasoning effort for every agent CLI

Set the reasoning effort once per CLI in Settings > AI providers and every Claude Code, Codex, Grok Build, Antigravity CLI or Kimi Code agent that has no effort of its own launches at that level. A project can override it, an agent or a ticket can pin its own, and AgentsRoom passes each CLI its own lever: --effort, model_reasoning_effort, --reasoning-effort or --thinking.

Learn more
Which model an agent starts on
  1. 1The model the agent, ticket or team step names
  2. 2The project's default model
  3. 3Default model in Settings > AI providers
  4. 4The CLI's own default

Switched off? The first model still on starts instead

Default model

A default model for every agent CLI

Pick a default model once per CLI in Settings > AI providers and every Claude Code, Codex, Antigravity CLI or other agent, ticket and team step left on Default starts on it. A project can use its own, a model you switched off under Models available to agents is never launched, and an agent that names a model keeps it.

Learn more
AgentsRoom unread indicators: WhatsApp-style badges showing which AI agents have new messages
Notifications

Know exactly which agent finished

Unread indicators per project and per agent, like a messaging app. Desktop and mobile push notifications when an agent completes or needs input. No more guessing.

Learn more
AgentsRoom Claude Code token usage : per-session token meter and session monitor with cache hit rate, input tokens, output tokens, cache reads and cache writes
Token Meter

Per-session Claude Code token usage, live, with cache hit rate

Every Claude agent terminal in AgentsRoom shows a live Claude Code token usage meter. The badge turns orange, then red, as the session's context window fills up. Click it to open the session monitor : input tokens, output tokens, cache writes, cache reads, cache hit rate, message count, models routed, session duration, plus contextual tips to lower your token consumption. All local, read straight from ~/.claude/projects/ JSONL.

Learn more
AgentsRoom project statistics: time spent per project, prompts, time per agent, tokens and cost, with a per-day chart and a per-agent breakdown
Project management

Project statistics: time, prompts and cost per project

Open the project statistics panel on any project and read time open, active and idle time, average active agents, prompts, agents opened, tokens and cost. A per-day activity chart and a per-agent breakdown table turn raw agent work into project analytics for billing, retros and cost control.

Learn more
Usage panel
Claude CodeCurrent session32%
Claude CodeCurrent week (all models)70%
Codex CLI5-hour window58%
Kimi CodeMonthly14%
Claude · AgentsRoom

Current week (all models) is at 76% (alert set at 75%). Resets Sep 24, 04:00

Also on your phone
Sent once per window, rearmed at the reset.No spam at every refresh
Usage alerts

A notification before the usage limit, not after

AgentsRoom reads the quota bars Claude Code, Codex, GitHub Copilot CLI, Grok Build, Kimi Code, Cursor, Antigravity, OpenCode Go, the Z.AI Coding Plan, Command Code and Muse Code publish, session, weekly, per model or per limit group, for every account you are signed in to, and puts a bell next to each one. When a bar crosses a threshold, 50% and then 75% by default, you get a native desktop notification, a push in the AgentsRoom mobile app, or both. Right-click a bell to pick thresholds for that bar or mute it. The reminder is sent once per quota window and rearmed at the reset, so an 80% bar is one banner, not one every refresh.

Learn more
AgentsRoom quota kick-off: a Claude Code or Codex session kicked off at 07:00 refills its 5-hour limit two hours sooner
Quota kick-off

Your 5-hour limit starts when you say

The countdown of a Claude, Codex or Antigravity session starts with the first message. Quota kick-off sends that message earlier, at set times or back to back, on the cheapest model, with no agent and no tab. Kicked off at 07:00, a session you start at 09:00 refills at 12:00 instead of 14:00.

Learn more
04Review & quality

Review and validate the work

AgentsRoom Commit Context: the AI agent conversation behind each commit is attached as an unlisted gist and linked from the commit message so any agent can recover the context
Commit Context · Agent memory

Every commit keeps the conversation that wrote it

AI agents have no memory between sessions, so the reasoning behind a commit usually dies when the terminal closes. Commit Context attaches the full agent conversation behind each commit as an unlisted gist and adds an ia-agents-conversation-history link to the commit message. The next agent or teammate opens that link and recovers the why behind the change: the prompts, the decisions, the constraints, the edits. Context is the value an agent needs to be precise, so AgentsRoom keeps it attached to your git history. Secrets are redacted before upload, the link is unlisted, and you can also create gists by hand on gist.agentsroom.dev. Works with Claude Code, Codex, GitHub Copilot CLI, Cursor and 15 other agent CLIs.

Learn more
AI Commit Messages
Assistant
fix temp tags
git diff3 files
Describe the commit, or send empty
AI Commit Messages

Commit messages written from your real diff

Stop typing "wip". Click the sparkle on the commit field and AgentsRoom reads the git diff of the files you are committing, then writes one clean message in Conventional Commits format. Type a rough draft like "fix temp pdf", get it back as a proper commit message. Your team's format rules, set once per project, are applied on every generation.

Learn more
AgentsRoom per-agent review: filter code changes by individual AI agent with scoped commit tabs
Multi-Agent

Diff per agent & task

When multiple agents work in parallel, filter code changes by agent. See only one agent's modifications, review their diffs in isolation, and commit each contribution separately. Clean git history, zero confusion.

Learn more
AgentsRoom Browser Automation: AI agents like the QA Engineer drive a real Chromium browser through MCP to test web apps end-to-end
Browser Automation

Your agents click, type and test the browser

AgentsRoom ships a built-in browser inside every project, plus an AgentsRoom Browser MCP server. Your QA agent navigates your localhost site, clicks, types, takes screenshots, reads console logs and verifies the feature is actually working before handing off. Real browser automation for Claude Code, no Playwright config to write.

Learn more
Plan review

Review the plan before the code

When an AI agent proposes a plan, AgentsRoom opens it in the agent's pane as a document. Select a passage to comment on it, replace it, delete it or add a step after it, then approve the plan or request changes in one message. Every version is kept with a diff, on desktop and on your phone. Claude Code plan mode, Codex Plan mode, the OpenCode Plan agent and Antigravity's implementation plan.

Learn more
One block, two readings
In the terminal
```mermaid
flowchart LR
  A[Login] --> B{Token valid?}
  B -->|yes| C[Dashboard]
  B -->|no| A
```
In AgentsRoomShow code
Mermaid diagrams

Mermaid diagrams, drawn instead of printed

Ask an agent for a diagram and it writes Mermaid, a block of text a terminal can only print. AgentsRoom draws every mermaid block wherever you read an agent: the conversation view, Markdown files and file tabs, Plan Review, notes, project memory and backlog tickets. Show code brings the source back, and a block that does not parse stays readable as code. Claude Code, Codex and every other agent CLI.

Learn more
One call, one page
The agent's terminal
view_present(
  path: "reports/q3.html",
  title: "Q3 dashboard"
)
The agent's paneDocuments3
Q3 dashboardq3.html · just now
Message the agent
Agent artifacts

Artifacts for every agent CLI

Your agent writes a dashboard, a chart or a report and shows it to you: a block above its composer opens the HTML page in one click, and a push opens it on your phone. The same tool for Claude Code, Codex and the other CLIs with an MCP client. Nothing is hosted: the page stays a file on your computer.

Learn more
AgentsRoom review mode: side-by-side file diff viewer for vibe coding with Claude Code
Vibe Coding

Review all changes, file by file

Navigate every file modified by your AI agents with arrow keys. See side-by-side diffs with syntax highlighting. Ask Claude questions about any change. Commit when you are ready.

Learn more
05Dev environment

Your dev environment

AgentsRoom localhost tunnel: expose local dev server as a public HTTPS URL with custom subdomain, free ngrok alternative
Tunnel

Share localhost with a public HTTPS URL

Expose your local dev server to the internet instantly. Pick a custom subdomain, get a public HTTPS link, and share your work with anyone, no deploy needed. Free ngrok alternative built into AgentsRoom.

Learn more
Chrome control
QA agent
checks the staging deploy
browser_select_target "Chrome · Pro"
Dev agent
updates the ticket
browser_select_target "Chrome · Personal"
Encrypted relay
Chrome · Prothis Mac
AgentsRoom started debugging this browser
Chrome · PersonalPC at home
browser_navigatebrowser_clickbrowser_screenshot
Dev environment · Chrome control

Your agents drive your real Chrome, one target per account

The AgentsRoom extension turns each Chrome profile into a named browser your agents can drive: Pro, Personal, Client, each with its own cookies and signed-in accounts. An agent picks its target by name, opens a fresh tab there and clicks, types, reads the console and takes screenshots, while another agent works on another account at the same time. The Chrome can sit on another machine of yours: the extension talks to the AgentsRoom relay, end-to-end encrypted, the same transport the mobile app uses. It is an MCP server, so Claude Code, Codex, GitHub Copilot CLI, Cursor and 15 other agent CLIs get the same tools. The embedded browser stays the default target.

Learn more
AgentsRoom terminal manager with detached window on a second screen, remote command launch from mobile, alternative to iTerm2, Warp and Ghostty
Terminal • Detachable • Mobile

A project terminal you drive from your pocket

A real per-project terminal manager built into AgentsRoom. Remember every command (backend, frontend, mobile bundler, Docker, worker, database) and launch them in one click. Detach the window onto a second monitor, or trigger commands from mobile over 4G/5G : your computer becomes a dev server you drive remotely. The modern alternative to iTerm2, Warp, Ghostty, Kitty, Alacritty, WezTerm, Hyper, Tabby and Terminal.app, designed for project work.

Learn more
Command palette⌘K
BacklogFix the login redirect loop#248
PromptsAudit a login flow
Filessrc/auth/login-form.tsxsrc/auth
Commitsfix: keep the login form focuseda3f19c2
↑↓move↵openescclose
Command Palette

One search bar for everything in your project

AgentsRoom holds your agents, your skills, your prompts, your scheduled agents and webhook triggers, your backlog, your dev commands, your SSH and database connections, and the repository itself. The command palette searches all of it from one field, opened with Cmd K on macOS or Ctrl K on Windows and Linux. Matching is fuzzy, so missing letters and missing accents still find what you meant, and results come back grouped by category instead of one flat list. A prefix character narrows a search to a single source, the arrow keys move, Enter opens, Escape puts you back where you were. Files honour .gitignore, commits come from the branch you are on, and the index is built in the background the first time you open the palette, never while the app is launching. Nothing you type is sent to a server.

Learn more
AgentsRoom restore agents and terminals on quit dialog : restore your AI agents, your dev servers and your terminal commands on the next launch
Restore • Resume

Restore agents and terminals on next launch

Quit AgentsRoom at the end of the day, choose to restore agents and terminals tomorrow, and reopen the app the next morning to find every AI agent, every dev terminal, every long-running dev server and every working directory exactly where you left them. One checkbox on quit, one click on launch, zero friction. Pick up where you left off and start coding in seconds instead of rebuilding your environment for an hour.

Learn more
AgentsRoom MCP : agent-pilotable IDE with Backlog MCP, Terminal Commands MCP, Prompt Library MCP and Browser MCP exposed to Claude, Codex, GitHub Copilot CLI, Cursor and 15 other agent CLIs
AgentsRoom MCP

An IDE your AI agents can pilot, not just chat with

Four MCP servers (Backlog MCP, Terminal Commands MCP, Prompt Library MCP, Browser MCP) expose the AgentsRoom IDE itself to Claude, Codex, GitHub Copilot CLI, Cursor and 15 other agent CLIs. Agents can read and update the backlog, reply to public backlog tickets, run saved dev commands (and create new ones), reuse or save prompts, and drive the embedded browser to verify their work. Manual when you want, automated when you don't. The unique edge that turns AgentsRoom into an agent-controllable IDE.

Learn more
Remote SSH
Encrypted SSH tunnel
AgentsRoom
prod-vps.acme.dev
staging.acme.dev
gpu-01.lab
SSH key:22
Remote server
deploy@prod-vps.acme.dev
Production VPS
Agent running
Save once, connect in one clickCredentials stay on your machine
SSH · Remote server

Run your agents straight on a remote server, over SSH

AgentsRoom is an SSH connection manager and a built-in SSH terminal in one. Save your VPS and remote servers once, connect with an SSH key or a password, then run Claude Code, Codex or Antigravity CLI directly on the remote machine, no separate SSH client to open. Pin a connection per project so a project that lives on a remote box always opens its terminal over the right SSH connection. Run heavy builds, GPU jobs or long-lived agents on a powerful server while you keep the same cockpit, status tracking and notifications on your laptop.

Learn more
Secret vault
Encrypted by the OS keychain
AI agent
STRIPE_SECRET_KEY
SENTRY_AUTH_TOKEN
NPM_TOKEN
Reads the names
Never receives a value
Machine-local vault
STRIPE_SECRET_KEY
0600, owner only
Dev server command
Resolved at launch : Main process only
Reference by name, resolve at launchNever leaves your machine
Dev environment · Secret manager

Secrets your agents can use but never read

API keys, tokens and passwords live in a machine-local vault encrypted through your OS keychain, in a single file written owner-only and atomically. You reference a secret by name, and AgentsRoom resolves it at launch, in the main process, into the environment of the command being started. An agent can list the names and ask for one to be injected, and there is no code path that returns a value to an agent, to the interface or over MCP. Nothing is synced to a server, and a machine without a usable keychain gets a refusal rather than a file in clear text.

Learn more
Database client
Read-only
AgentsRoom
SELECT id, email, plan FROM users LIMIT 50
DELETE FROM users WHERE id = 42
Blocked : connection is read-only
SSH tunnelbastion.acme.dev
Private subnet
Capped result set
Through your saved connection
One statement per callPassword never leaves the app
Dev environment · Database

Let agents query your databases, read-only by default

Save your MySQL, PostgreSQL, SQL Server, MongoDB and SQLite connections with host, port, user, database and TLS, and your Google BigQuery projects with your own Google login, then reach the ones that are not publicly exposed through an SSH tunnel or an AWS SSM port-forwarding session built on the connections you already saved. Every connection starts read-only : a write needs both a writable connection and an explicit confirmation, and a connection flagged as production makes that confirmation harder. One statement per call, stacked statements refused, result sets capped. Agents query over MCP by naming a connection, so the password never reaches them. Your connections follow your account from one machine to the next, and a database can be exported to a file, imported back, or copied straight onto another server, with a backup of the target taken first.

Learn more
SSM port forwarding
No public endpoint
AgentsRoom
127.0.0.1 : port picked by the OS
aws ssm start-session
AWS-StartPortForwardingSessionToRemoteHost
Managed nodei-0a1b2c3d4e5f
RDS in a private subnet
private subnet : 3306
Reserving a free loopback port
Loopback only, never the local networkNo inbound SSH, no public endpoint
Dev environment · AWS SSM

Reach an RDS database in a private subnet, without a bastion

Your Amazon RDS instance has no public endpoint, no inbound SSH and a security group that only lets the application in. AgentsRoom reaches it through an AWS Systems Manager port-forwarding session : it asks the operating system for a free loopback port, runs aws ssm start-session with the AWS-StartPortForwardingSessionToRemoteHost document through a managed EC2 node, and waits until that port really accepts a TCP connection before connecting its database client. No bastion host to patch, no VPN to provision, no endpoint and password copied into a chat message. Authorization comes from the AWS profile already configured on your machine, so no key is stored by the app, and the forwarded port is bound to 127.0.0.1 only. MySQL, PostgreSQL and MongoDB, so RDS MySQL, RDS PostgreSQL, Aurora and DocumentDB.

Learn more
Cloud SQL Auth Proxy
No public IP
AgentsRoom
127.0.0.1 : port picked by the OS
cloud-sql-proxy
--private-ip --auto-iam-authn
Your gcloud loginTLS 1.3, IAM
Cloud SQL instance
acme-prod:europe-west1:orders
Reserving a free loopback port
Loopback only, never the local networkNo key stored, no bastion VM
Dev environment · Google Cloud

Reach a Cloud SQL instance on its private IP, without a bastion VM

Your Cloud SQL for MySQL or PostgreSQL instance has no public IP, and nobody wants a bastion VM just for the occasional query. Pick the instance by its connection name, or from the list your gcloud login can see, and AgentsRoom starts the Cloud SQL Auth Proxy on a free loopback port when the connection opens, then stops it when it closes. It runs with your Application Default Credentials, so no service account key is stored, and IAM database authentication removes the database password too. When the proxy is missing or your login has expired, the error says so and names the command to run. AI agents query the same instance read-only, without ever seeing a credential.

Learn more
One exit per provider
  • Claude CodeCompany proxyproxy.corp.example:3128Exit IP203.0.113.10
  • CodexHome proxyhome.example.net:8080Exit IP198.51.100.24
  • GitHub Copilot CLIDirectExit IP192.0.2.7
Dev environment · Network proxy

Give each agent CLI its own network proxy

Exporting HTTPS_PROXY in your shell sends every CLI through the same exit. In Settings > AI providers, each provider has its own Proxy field instead: Claude Code can leave through the company proxy while Codex uses your home connection, at the same time. Every agent of that provider starts with HTTPS_PROXY, HTTP_PROXY, ALL_PROXY and NO_PROXY set, AgentsRoom's own quota checks for that provider take the same route, and Test shows the exit IP before you launch anything. A proxy password is a reference to the Secret Manager, never plain text.

Learn more
What still stops an agent
  • Bypass Permissions mode warningAnswered by AgentsRoom
  • Yes, I trust this folderAnswered by AgentsRoom
  • Dangerous rm operationAnswered by AgentsRoom
  • Input holds your account emailAnswered by AgentsRoom
Dev environment · Agent permissions

Skip permission prompts, and what still asks

Choose Autonomous or Manual once per agent CLI. Autonomous launches Claude Code, Codex and the others with their own bypass flag, then answers what Claude Code keeps asking: the Bypass warning, Yes, I trust this folder, ordinary deletions.

Learn more
One list, every CLI
Your MCP servers
  • notionmcp.notion.com/mcpConnected on this computer
  • sentrymcp.sentry.dev/mcpRemote
  • githubnpx github-mcpLocal
  • Claude Code
  • Codex
  • GitHub Copilot CLI
  • Cursor
  • and 14 other agent CLIs
Dev environment · MCP servers

Hand the same MCP servers to every agent CLI

Claude Code, Codex and every other CLI keep their MCP servers in their own file, with their own login. In Settings > Agent behavior > MCP servers, you declare a server once, on every agent, a project or one agent, and AgentsRoom writes it into each CLI's configuration at launch. Add it from the catalog of the official MCP registry or paste the mcpServers JSON you already use, keep its key in the Secret Manager, and connect an OAuth server like Notion, Linear or Atlassian once per computer with the Connect button.

Learn more
Harness of one agent
PersonaCode reviewer
ToolkitResearch
  • scan-documents
  • summarize
  • notes (MCP)
  • staging-db
Project memoryRead only
At launch
  • Claude Code
  • Codex
  • Antigravity
  • Cursor
Dev environment · Toolkits

Harness engineering: persona, tools and memory as three controls

A toolkit bundles skills, MCP servers, saved SSH, database and dev command connections, and environment variables. Attach it by reference to an agent, a single backlog ticket, a team step, a project or a client folder, and Claude Code, Codex and every other agent CLI get it at launch. Project memory is a separate setting per agent and per ticket: off, read only, or read and write, without touching the persona.

Learn more
deploy.sh

$ curl POST /api/dev-inboxes/…/push

{"title":"Deploy finished"}

202 Accepted

Your phonenow

Deploy finished

api v2.4 is live in production

Push API

Push notifications from cURL

Send a push notification to your iPhone or Android phone with one cURL call from a deploy script, a CI job or a cron task. Send-only revocable keys, no duplicates on retries, works with your computer off.

Learn more
Project inboxbilling-api-3f9c…@inbox.example.com
Search messages
  • cron@billing-api06:00

    Nightly invoicing report

    412 invoices sent, 0 failed

  • [email protected]2 min

    Checkout API is down

    HTTP 503 from 3 regions since 09:42

Email inbox

A developer email inbox per project

Give each project its own receiving email address for service alerts, cron reports and app emails. Plain-text reading on desktop and mobile, revocable address, optional sharing with your team.

Learn more
CLI Doctor
Launch failed
$ claude --model opus-5 --mcp-debug
error: unknown option '--mcp-debug'
Dev environment · CLI Doctor

Know why your agent CLI would not start

An agent CLI that dies at launch usually says nothing useful, and the cause is buried in a command you never typed. AgentsRoom records the exact command it ran, tags every flag with the setting that injected it, reads what the CLI printed, and answers in plain language: what broke, and the ordered steps to fix it. Deterministic checks like a missing binary or a rejected flag are answered instantly without any AI call, redaction happens on your machine, and nothing is ever executed for you.

Learn more
Repository import
ConnectGitHubGitLabor your own GitLab host
Clone and add as projects
Dev environment · Repository import

Bring your GitHub and GitLab repositories in

Connect GitHub or GitLab once, see the repositories you actually have access to, tick the ones you want and clone them in a single pass. Each one lands as a project with its path already set, ready to receive agents. Authenticate with a token or reuse the gh or glab session already signed in on your machine, point at your own self-hosted GitLab host if you have one, and keep everything local: the token is encrypted on disk and clone credentials are never written into your git config.

Learn more
06Anywhere

Anywhere, with any agent

AgentsRoom mobile-desktop sync: control AI coding agents from iPhone and Android with encrypted real-time sync
Mobile Companion

Control your agents from your phone

Start agents, type in the terminal, send prompts, and monitor progress from your iPhone or Android. Real-time sync with end-to-end encryption between your desktop and your mobile device.

Learn more
Lock screenWorking

3 agents working

  • BackendRate limiting on the API12 min
  • FrontendDark mode for settings7 min
  • QACheckout end-to-end tests3 min
One Live Activity for the whole roomEnd-to-end encrypted
Live agents

Every agent you run, on your lock screen

One Live Activity for your whole room of AI coding agents, whatever the CLI and whatever the computer: a status ring with each agent, its task and a live timer, on the iPhone lock screen, in the Dynamic Island, on Apple Watch and in StandBy. When an agent needs you, its question comes first, and on iPhone that is the only update that plays a sound. Answer a Claude Code question or permission request from the lock screen, behind Face ID. When the room goes quiet, a summary card shows the duration, the agents, the files and the lines changed. Names, tasks and questions are end-to-end encrypted. Home and lock screen widgets, a quota widget, and a live notification with a Live Update on Android. Free for everyone.

Learn more
AgentsRoom multi-provider: switch between Claude, Codex, GitHub Copilot CLI, Cursor and 15 other agent CLIs while keeping context
Multi-Provider

Switch between Claude, Codex, GitHub Copilot CLI, Cursor & 15 other agent CLIs

Use 19 AI coding providers in the same project. Pick a provider per agent, switch mid-conversation, and keep your full context with automatic handoff. Claude, Codex, GitHub Copilot CLI, OpenCode, Antigravity CLI, Aider, Grok Build, Mistral Vibe, Kimi Code, Amp, oh-my-pi, Freebuff, Devin, Cursor, Command Code, Muse Code, Pi, MiniMax Code, ZCode: all in one app.

Learn more
AgentsRoom Claude multi-account: pick Work, Personal or client Claude Code account per project and per agent, sign-in built into the IDE
Multi-Account

Several Claude Code accounts, one app, zero shell gymnastics

Use a Work Claude account, a Personal one, a client account, all at the same time in AgentsRoom. Pin one account per project, override per agent. Sign in directly from the Settings panel, no terminal trick required. Compatible with CCS profiles too.

Learn more
AgentsRoom Codex multi-account: Multiple Accounts panel with Claude and Codex tabs, showing per-project and per-agent Codex account picker with sign-in status badges
Codex Multi-Account

Several Codex accounts, one app, zero shell gymnastics

Use a Work Codex account, a Personal one, a client account, all at the same time in AgentsRoom. Pin one account per project, override per agent. Sign in via codex login directly from the Settings panel, no terminal trick required.

Learn more
Grok Multi-AccountGROK_HOME
Work account
~/.grok-work
Signed in
Agent running
Personal account
~/.grok
Signed in
Agent running
Two signed-in accounts, two agents, one window
Grok Multi-Account

Several Grok accounts, one app, zero logout

Use a work Grok Build account, a personal one and a client account at the same time in AgentsRoom. Pin one account per project or per folder, override per agent. Sign in with grok login straight from the Settings panel, no GROK_HOME to export by hand.

Learn more
Cursor Multi-AccountCURSOR_CONFIG_DIR
Work account
~/.cursor-work
Signed in
Agent running
Personal account
~/.cursor
Signed in
Agent running
Two signed-in accounts, two agents, one window
Cursor Multi-Account

Several Cursor accounts, one app, zero logout

Use a work Cursor account, a personal one and a client account at the same time in AgentsRoom. Pin one account per project or per folder, override per agent. Sign in with cursor-agent login straight from the Settings panel, no CURSOR_CONFIG_DIR to export by hand.

Learn more
Auto-switch
Working
Account 1
78%
Account 2
12%
Same conversation, same taskContext kept
Account auto-switch

Hit a usage limit, keep working

When the Claude, Codex or Grok Build account an agent runs on reaches the quota threshold you set, a quota rule hands the work to another account you are already signed in to and it carries on: the same session with its full context, no restart and no waiting for the 5-hour or weekly window to reset. The exhausted account is benched until its quota comes back, then rejoins the pool by itself. It turns several smaller subscriptions into something that behaves like one much bigger plan, and it keeps unattended runs alive: a scheduled task at 2am no longer dies silently at the limit.

Learn more
Alternate providersClaude ×1, then Codex ×1
ClaudeIts turn
Running agent · keeps its tool
Codex
Next launchClaude
Claude and Codex together

Use Claude Code and Codex together, on both quotas

A quota rule shares your new agents between Claude Code and Codex, or between two Claude accounts, in turn: one ticket out of two, or 10% of the Claude session then 10% of the Codex week. Agents already running keep their tool, a full provider skips its turn, and the rule can be limited to the backlog tickets of one project. Other rules act at a threshold: switch to Codex when Claude is almost full, use a lighter model, block or redirect launches, run a trigger, or move running agents to another account.

Learn more
AgentsRoom Cloud Agents: the Run in the cloud window with the prompt, the Claude and Codex choice, the model row and the dedicated branch a disposable machine will push
Anywhere · Cloud agents

A task runs while your computer is off

Trigger a coding task from the desktop app or from your phone, and it runs on a disposable machine in the cloud instead of your laptop. A cloud agent is a job, not a remote session: the machine clones the repository, runs Claude or Codex headless, commits, pushes a dedicated branch and writes a report, then destroys itself. Since the meeting point is git, nothing on your side has to stay awake, so a user report can land on your public backlog, become a cloud run started from the backlog tab of your phone, and reach production through your own CI/CD without a single computer of yours being switched on. When the run ends you validate the branch, relaunch it in the cloud, or continue locally in a normal session with the context already filled in.

Learn more
Remote Fleet
End-to-end encrypted
Office desktop
Frontend agent
Running agent
Home laptop
Backend agent
Build server
Offline
One window, every machine
Streaming live from Office desktop
Live
Anywhere · Remote fleet

Every project, every machine, one window

You do not code on a single computer anymore. There is the office desktop, the laptop at home, maybe a build server in the corner. Remote Fleet links them into one unified view so every project and every Claude, Codex or Antigravity agent shows up in a single window, whichever machine it actually runs on. Each machine keeps running its agents locally, and the fleet view aggregates their status, terminals and projects. Cross-device orchestration travels over an exclusive end-to-end encrypted relay, so your code and your agent output never leave your machines in the clear.

Learn more
Continue on another computerWorking at the office
Office desktop
Backend DevPlan the API migrationWrite the new endpointsFix the two failing tests
Encrypted relay
Home laptop
Same conversation3 / 3
Anywhere · Move an agent

Continue the conversation on your other computer

Started with an agent at the office and want to keep going at home? Right-click it, pick your laptop, and it picks up the same conversation there, where it left off. Claude Code and Codex resume their exact conversation, sub-agents and tool outputs included; every other CLI starts a new session that first reads the transcript of its console. The console on the first computer closes, the conversation travels end-to-end encrypted through a relay room only your two computers join, and nothing is deleted on the way.

Learn more
Machine poolA new agent asks for a computer
New agent
  1. 1Always-on box
    6 / 6 agents
  2. 2Office PC
    2 / 6 agents
  3. 3Laptop
    0 / 3 agents
Anywhere · Machine pool

Agents start on the computer that has room

An always-on box and a laptop or an office PC: give a project an ordered list of computers, its machine pool, and every new agent starts on the first one that is online, has the project and is under its budget. Opening an agent, Start on a ticket, the quick agent and triggers set to the pool all follow it, and a machine you pick by hand still wins. Each computer gets a budget for all projects together: a maximum of live agents, a free memory floor and a stricter night cap. A project that lives on the other computer opens in full: files, changes with commit and push, backlog and terminals.

Learn more
AgentsRoom remote agent control: piloting a Claude, Codex or Antigravity agent running on a desktop computer from an iPhone, with live terminal output
Anywhere · Remote control

Take over a running agent from your phone

The agent keeps running on your desktop, through its own CLI, on your machine. Your phone becomes the remote: read the terminal as it streams, answer a permission prompt, send the next instruction, fire a saved prompt, move a backlog ticket, open a tunneled dev server in the phone browser. Push notifications tell you the moment an agent finishes or gets stuck waiting for you, so a long task no longer chains you to the desk. Everything travels over the AgentsRoom end-to-end encrypted relay.

Learn more