Autorun
AGI Labo membershipTasks that wake up
in the morning.
"Summarize today's news at 6 AM." Schedule by time or interval. Results appear in the same task list.
AGI CockpitThe work OS
for AI agents
Claude Code, Codex, Cursor, Grok Build — plus OpenRouter and LM Studio.
Hand over a rough request, and your AI team splits it up and gets moving.
From asking to approving, work starts and finishes here — in one screen.
Windows · macOS Apple Silicon · Linux · Local features free & unlimitedClaude Code, Codex, and other agents require their own plans
Supported agents
Cockpit Agentfirst-party
Claude Code
Codex
Antigravity
Cursor
Grok Build
Master Agent
A rough ask
is all it takes.
"Prep next week's launch." "Write the monthly report." Ask casually, and Master Agent decides which AI to hand it to, runs them in parallel when needed, and brings the results back together.
- 01 Read the request, figure out the steps
- 02 Run the needed AIs in parallel
- 03 Aggregate progress, ask you only when it needs to
One screen
Many AI agents. One screen.
What's running, what's waiting, what's done. Cockpit ranks them by what needs your eyes — so handing more work to AI doesn't mean losing track of it.
- Cockpit Agent— first-party · any model you like
- Claude Code
- Codex
- Antigravity
- Cursor
- Grok Build
- Terminal— your usual CLI, too
Statuses update on their own. The time you spend checking “is it still running?” goes away.
cockpit ask
When AI needs a call,
it asks you.
"Which one?" "Should I go ahead?" A floating window stays on top, and you reply with a tap or free text. You can answer from your phone, too. All you do is decide.
Native chat
What the AI did,
in a form you can read.
Chat-style responses. File changes. Inline previews. You can tell at a glance what happened. And while it's running, your next instructions just go on the queue.
Diffs
Changed files side by side
Preview
Articles, Markdown right there
Queue
Stack next instructions
Images
Generated images in-thread
HTML Mode
Answers you can
actually use.
Dashboards, choice buttons, forms. Instead of reading a paragraph and typing a reply, you look, click, and move on. Whatever you input flows back to the AI to continue.
PWAWorks from mobile
Cockpit Agent
Bring your own
model.
The first-party Cockpit Agent runs on cloud models via OpenRouter or local models via LM Studio — so you can balance cost, privacy, and quality per task.
OpenRouter
Pick from Grok, Qwen, DeepSeek and more, with price and context length right in the picker.
LM Studio
Use a locally-loaded model on your own machine, ideal for sensitive data.
Autonomy
Plan only, approve risky actions, or fully auto — dial in how much you hand over, anytime.
Built for daily use
Keep things moving
when you're away.
Schedule. Reply from your phone. Call up Cockpit from your usual Claude Code or Codex. Built for fitting AI into actual daily routines.
Tasks that wake up
PWA Mobile
AGI Labo membershipMove work forward
from your phone.
Create tasks, send follow-ups, attach images, and reply to cockpit ask — all from a PWA on the go.
cockpit skill
Spawn tasks
from a chat.
Install the cockpit skill into Claude Code, Codex, Antigravity, Cursor, or Grok Build. They can create Cockpit tasks right from a normal conversation.
Real tasks
What we hand over,
every day.
At AGI Labo, the whole company runs through Cockpit — development, marketing, even event ops. A few real tasks from a recent day.
“Create the Luma event for the next lecture, from thumbnail to publish”
Claude CodeDone
“Transcribe yesterday’s lecture recording and shape it into an archive article”
Claude CodeRunning
“Draft three X posts from the v4.17 release notes”
Cockpit AgentWaiting
“Triage open GitHub issues and propose today’s top three”
Autorun · weekday morningsDone
“Pull this week’s membership numbers from Stripe into a report”
Autorun · weeklyRunning
“Reorganize this page’s benefits and propose new hero copy”
Claude CodeDone
Writing code, drafting articles, crunching numbers, setting up events — if it happens on a computer, you can hand it over.
Voices
What users are saying.
From developers to founders — real posts from X.
AGI Cockpit is incredible. I can develop an iPhone app using only my iPhone: work on it from bed, spot a bug, request a fix on the spot, and have the corrected build launch on my phone three minutes later. I can even push the latest build to App Store Connect from my phone through Claude Code skills. The future is here.
No exaggeration: AGI Cockpit is genuinely excellent for moving development forward in parallel. It fits this new era of running multiple agents at once. When I’m building, I keep wanting to queue the next task, and Cockpit is much better suited to that than splitting a terminal into panes just to assign work.
Specs
At a glance
| Supported agents | Claude Code / Codex / Antigravity / Cursor / Grok Build / Cockpit Agent (first-party) |
|---|---|
| Model providers | OpenRouter and LM Studio (local models) |
| Platforms | Windows / macOS (Apple Silicon) / Linux (AppImage and .deb, x64 and arm64) |
| Pricing | Local features are free and unlimited, with no account required. AGI Labo membership is $20/month or $200/year, with a 7-day free trial |
| Free features | Every local feature, including parallel task management, Master Agent task splitting, HTML Mode, cockpit ask, the side panel, in-app browser, and session history dashboard |
| AGI Labo membership adds | Autorun scheduling and remote access from your phone or another device |
| Distribution | Microsoft Store / .dmg / AppImage and .deb |
FAQ
Frequently asked questions
- What else do I need?
To use Claude Code, Codex, and the like, you need each agent’s CLI and plan; the app guides you through installing any missing CLIs. The first-party Cockpit Agent runs with an OpenRouter API key or a local LM Studio.
- How much can I use for free?
Every local feature is free and unlimited. Parallel task management, every supported agent, Master Agent, HTML Mode, and the session history dashboard all work without an account, with no task caps or trial timer.
- What does AGI Labo membership add?
Exactly two Cockpit features: Autorun scheduling and remote access from your phone or another device. Membership is $20/month or $200/year with a 7-day free trial, and also includes AGI Labo articles, courses, and community benefits.
View AGI Labo membership plans- Which platforms are supported?
Windows, macOS (Apple Silicon), and Linux (AppImage / .deb, x64 and arm64).
- How is it different from other agent managers like Vibe Kanban or Conductor?
AGI Cockpit is a work OS, not just a coding-agent runner. It unifies Claude Code, Codex, Antigravity, Cursor, and Grok Build — plus OpenRouter and LM Studio models — on one screen, runs on Windows, macOS, and Linux, and adds Master Agent task splitting, Autorun scheduling, remote control from your phone, and HTML Mode for rich outputs. That makes it suited for any work on your computer, not only coding.
- How is it different from the official Claude Code desktop app?
The official app is dedicated to Claude Code. AGI Cockpit runs multiple agents side by side in one task list — including Claude Code — with task splitting, scheduled runs, and mobile access in a single screen. They also work together: install the cockpit skill in Claude Code and it can create Cockpit subtasks from your usual sessions.
- Do I need to be a developer?
No. Writing articles, crunching numbers, preparing events, research — anything you do on a computer can be delegated. At AGI Labo we run development, marketing, and event operations entirely on Cockpit.
- Norton quarantines the Windows version as “IDP.HELU.PSE79” during install
This is a false positive by Norton’s heuristic detection (we have filed a false-positive report with Norton). Open Norton’s Security → History, switch the view to Quarantine, select the AGI Cockpit item, and choose “Restore & exclude this file”. Then launch the app again; if it does not start, reinstall it from the Microsoft Store.
Release history
Releases
Since v2.0
- Added mobile PWA controls to check, download, and install host updates, with confirmation before restarting the host.
- Added /compact conversation context compaction for Antigravity, Cursor, and Qoder.
- Added model list sources, retrieval status, and a reload action.
- Added guidance for agent CLI updates needed for new models, with CLI updates available from Desktop.
- Improved Autorun validation to check that the selected model and reasoning effort are available before saving.
- Fixed viewport height and composer positioning after pairing or reconnecting in the PWA.
- Improved selected-text capture on macOS by waiting for held hotkey modifiers to be released before copying.
- Improved Cockpit Agent so new tasks and autoruns use the models and reasoning levels supported by the connected provider.
- Improved Codex so the latest models available to the selected account can be chosen in New Task, task settings, and autoruns.
- Improved cockpit task create and task run so they reconcile an already-created task after a lost response and avoid creating duplicates.
- Improved project deletion so you can delete all tasks in a project, including running tasks, after confirming the impact.
- Added links to each agent's provider status page when a service outage is detected.
- Improved the in-app Browser so covering overlays show a page snapshot instead of a blank white surface.
- Fixed Antigravity Visual so High reasoning effort is kept after a turn completes instead of snapping back to Low.
- Improved Cockpit Ask so JSON summaries with choices are converted into structured Asks instead of showing raw JSON.
- Fixed Codex usage and rate-limit retrieval for Codex CLI 0.153.
Download
Start by handing it
one thing.
Morning news digest. A draft article. Cleaning up notes. Put one thing you already do every day into Cockpit, and the rest falls into place.
Windows · macOS Apple Silicon · Linux · Local features free & unlimitedClaude Code, Codex, and other agents require their own plans