intentic · An IDE for your agents. A window for you. Product Run agents<br>Fleet board<br>Chat & plan mode<br>Review & land<br>The environment<br>Workspace & editor<br>Capabilities<br>Sandbox & ownership<br>Extend it<br>Doorbell<br>Extension gallery<br>Try the live workspace →
Docs Get started<br>Overview<br>Quickstart<br>Docker setup<br>Understand<br>Architecture<br>Extend<br>Extensions<br>Build an extension<br>Manifest reference<br>Publish & marketplace<br>Guides<br>Doorbell<br>Autonomous employees<br>Reference architecture<br>Release notes (opens in a new tab)
Extensions<br>Compare<br>Download<br>About
GitLab
Get started free
An IDE for your agents.<br>A window for you.<br>Every agent gets its own sandbox and git worktree, on hardware you own. Run ten at once. Nothing lands until you have read the diff.<br>Get started free See the source<br>Free plan<br>Bring your own agent<br>Runs on your hardware
acme-shop · /agents<br>Try the live workspace
The real app on a recorded workspace. Approve a plan, answer an agent, read a diff. Nothing to install.
The loop<br>Run several at once. Land them one at a time.<br>The parallel half is safe by construction: one sandbox and one worktree each. The deciding half stays yours. Works with Claude Code, Codex, Grok, Kimi Code and Gemini.<br>01<br>Give the work to as many agents as it needs<br>Each starts in plan mode on a git worktree cut from your base commit. The board sorts them by who needs you; the rest keep running while you answer.
acme-shop · /agents
02<br>Each works in a sandbox of its own<br>Not a chat window: a container on your machine, with the job's toolchain really installed and its context loaded every turn. It never touches your working tree.
your machine · reached from your browser over a private tunnel<br>release-captain<br>owns the weekly release
running
Environment · installed<br>node 24 pnpm docker psql
Context · loaded every turn<br>6 skills release runbook house style
03<br>Nothing lands until you have read the diff<br>Finished work arrives as a branch: every changed file, every hunk, the tests it ran. Land it as ordinary git changes, or discard it and the worktree goes too.
acme-shop · agent/soft-deletes
Every surface, with the screenshots →
Ownership<br>Your code never leaves your machine.<br>An agent with your keys is only safe if you own where it runs. Here is exactly what sits on each side.<br>Your machine<br>Where the sandbox runs.<br>Your repositories and working tree<br>API keys, .env files, database passwords<br>The agents, their sandboxes, their history<br>Every file an agent reads or writes
The intentic platform<br>Everything it stores, in full.<br>Your email address<br>Your sandbox's URL<br>No code, no keys, and no way to command your agents.
Your browser holds the token that drives the sandbox; the platform never does. The sandbox is MIT on GitLab, so you can check.<br>How ownership works in the product →
Economics<br>A whole fleet, on the subscriptions you already pay for.<br>Ten agents sounds expensive. It isn't. Each runs on a subscription you already pay for, on hardware you already own.<br>Your agent accounts<br>Claude<br>Opus, Sonnet and Haiku, on your Claude plan<br>subscription<br>Codex<br>on your ChatGPT plan<br>subscription<br>Grok<br>on your SuperGrok plan<br>subscription<br>Kimi Code<br>on your Kimi Membership<br>subscription<br>Gemini<br>on your Google account<br>subscription
Flat, not metered<br>No per-token metering. No markup on your model usage.<br>No rented cloud compute. Agents run where you run them.<br>Free for one sandbox; Pro unlocks the fleet and sharing.
Extend it<br>A small core. Everything else is an extension.<br>The workspace ships lean: the fleet, the sandbox, the editor, the review. Everything past that is an extension, allowed to do only what its manifest declares.<br>Automations
Agents that start themselves, so nobody has to be at a keyboard. Each run is a fresh session with its own transcript, gated by a guard command you write, and one agent's run can wake the next.<br>Wakes on<br>a push<br>a Sentry alert<br>a Stripe payment<br>inbound email<br>a chat message<br>plain cron
Discord & Slack
Assign work with an @mention. It replies in the thread, with receipts.<br>Doorbell
Put an agent on your own website behind one script tag.<br>Team sharing
Invite teammates into one sandbox, each over their own tunnel.<br>Memory & pipelines
Notes that outlive a run, CI it can fix, a preview per repo.<br>A whole company
One agent per role and per team, sharing the services they all use.
An extension adds tools, skills and image layers, not just UI. Yours stays in your own repo.<br>Browse every extension →
About the creator<br>Artur Kurowski<br>Full-stack engineer, DevOps, and 15+ years of shipping production systems.<br>I built intentic because I wanted a real workspace for my own agents. Not a chat window, and not somebody else's cloud holding my source. It runs on my hardware, and now on yours.<br>GitHub<br>LinkedIn<br>radarsu.com<br>More about who builds this →<br>Why trust intentic?
Built by a verifiable person, and built so that trusting the person is optional. The platform holds your identity and a URL, with no path to your code, your keys, or your sandbox.<br>What the platform actually holds...