remoteagent.online · open source

An AI agent that runs on your machine.

Every dream starts as a sentence. Give it a real machine, your real files, and a record that cannot be rewritten afterwards — and it stops being a maybe.

The cloud queues what it may do. Your device decides what actually happens.

installchecksum-verified
$ curl -fsSL https://remoteagent.online/install.sh | bash $ remote-agent login # connect this device with your Remote Agent key $ remote-agent gui # or: remote-agent start (headless)
20 built-in tools 232 commands · 9 categories 6 device skills 5 LLM providers open source
Where does it actually run?

A Node daemon (remote-agent) on your machine, plus the cloud console. Your files never move: the cloud queues typed operations and your device executes them.

What is open source?

The device client is public at github.com/airemoteagent/AI-Remote-Agent; the server side is the mona-plugin (1.2.0) in this app. You can read both.

What do I need to start?

One LLM provider key — OpenAI, Anthropic, Google, DeepSeek or OpenRouter — and a machine to run the agent on. The key is stored encrypted on the server, never on the device.

In plain words

No jargon. Six answers.

What is it?A helper that does computer work for you — and writes down exactly what it did.
Where does it run?On your computer. Windows, Mac, Linux, or a Raspberry Pi in the corner.
Does it touch my files?Only the folders you tick. Everything else is out of reach.
What does it cost?You pay your AI provider. Here you see every run's cost. A spending cap is not built yet — no pretending.
Can I stop it?Yes. Stop the agent with one click, or switch the device off. Nothing runs without it.
Do I need to be technical?No. One command to install, then you just type what you want in plain English.

How it works

Queued in the cloud. Decided on your machine.

One pipeline for every action. Nothing skips the gate in the middle.

Intentyou ask
Policymay it?
Your deviceit runs here
Recordedsteps, tokens, cost
Runs on your device, not ours

The daemon polls for work and claims each task atomically, so only one of your machines executes it. The console shows which device is online and what it is running.

A manifest, never your files

Workspace sync sends labels, counts and an identity hash — no host paths, no file bodies. Read, write, patch, rename, copy, trash and diff all run locally on the device.

Policy before every tool call

Rules are first-match and deny by default. On POSIX, shell commands are argv-only with an allow-list and a scrubbed environment; risky patterns can require your approval; requests are rate-limited.

Network access is closed by default

Outbound requests resolve DNS first and refuse loopback, private, link-local, metadata and CGNAT ranges. Redirects are validated again, with 50 KB and 15-second caps.

Durable runs, not best-effort

A run moves created → running → verifying → succeeded / failed / cancelled with per-step checkpoints. A crash resumes instead of repeating, and side-effecting retries require an idempotency key.

The world is data, never instructions

Tool output, files and web pages are labelled untrusted in the agent's own prompt and are never followed as commands.

Limits that are enforced

Plan limits are checked server-side: agents, runs per day, steps per task and requests per minute. Free is 10 agents · 200 runs/day · 8 steps · 60/min; Pro raises all four.

Keys you never see again

Provider keys are stored AES-256-GCM encrypted. The API returns them masked, never in full, and sensitive actions ask for a second factor first.

Evidence you can check

The device keeps a hash-chained local audit log and anchors its chain head to the cloud (append-only: sequence, hash, signature). The console keeps the per-step trace of every run.

The console

One console for agents, files, runs and cost.

Click through it — this is the real console, panel by panel.

remoteagent.online / control room

Get your first agent running in 3 steps

  1. Install the agent on your device
  2. Connect it with your Remote Agent key
  3. Send a task — watch it stream back live
RunningAgentsDevicesTasks todaySpend (7d)Commands

The real dashboard keeps these live: agent status, device telemetry, runs and spend.

Real files. Visible changes. Verified delivery.

index.htmlstyles.cssapp.jsassets/

<!doctype html> <html lang="en"> <head> <meta charset="utf-8"> <title>My app</title> </head> <body> … </body> </html>

Acceptance

Renders, no console errors

Evidence

Verdict recorded against the run

content hash recorded on every change · versions kept

Your files stay on your machine — the cloud only queues the operations your device runs.

Your helpers and their rules

Assistant● runningsecure modeallowlist shell · audited
deepseek/deepseek-v4
MacBook · online
ls · cat · grep · git · node … or *
Desktop · Documents · Downloads · Home · Workspace

20 built-in tools, each one yours to switch off

sysinfoenvsessionsettingsdiscovershellfilesnetappsbrowserwebmemorynotifyvectorjobsdelegategoalworkflowpluginrecall

Receipts: steps, tokens, cost

donedeepseek-chat9,947 tokens≈ $0.0028

Example 1 from the published docs — a real captured run.

think — the user wants Safari, then YouTube… tool.call apps — launch Safari tool.result exitCode 0 — Safari is up tool.call browser — open youtube.com tool.result opened: true answer Done — Safari is open on YouTube verify checked the tool results

Brain mode for this run — the console compares them from real runs:

autocheapbalanceddeepmanual

What it cost — per day, model and agent

Google Maps clone · 24 steps · 76.5k tokens · $0.04
App Store clone · 30 steps · 48.2k tokens · $0.02
Chess game · 9 steps · 24.2k tokens · $0.0079

The three published one-prompt builds — real captured cost, drawn to scale.

cost per daytokens per daytasks per dayby providerby modelby agent

What waits for your OK

shell command beyond the allow-listgit push --forceexample request
ApproveDeny

Approve or deny — the decision is recorded with the request, and denying asks for a note.

The console's own empty state: “No pending approvals — everything is settled…”

Skills and the command allow-list

briefingdisk-healthnetwork-healthservice-healthweb-researchcertificate-expiry

6 built-in skills — written procedures the agent follows, switched on or off per agent.

232 commands in 9 real categories — the allow-list you pick from, or allow everything with *.

files 47system 47dev 39util 25network 23process 18packages 14media 11power 8

Custom commands (dynamic tools) you add are pushed to your devices automatically.

Everything else, organised

Free — 10 agents · 200 runs/day · 8 steps · 60 req/min
mode · max steps 2–16 · temperature · answer verification · extra rules
OpenAI · Anthropic · Google · DeepSeek · OpenRouter — AES-256-GCM
2FA — authenticator, backup codes, email code
persona · session memory · rolling summaries · approval policy
factory reset (type RESET)

Plan limits are enforced server-side. A user-set spending cap is not built yet — the console says so here.

Click any section. Layouts are the real console; anything illustrative is labelled, and every figure shown is real.

Workspaces

File tree and editor, diff, sandboxed preview with start/stop, versioned changes with content hashes, acceptance criteria, evidence with a verdict, agents granted review-only or read/write access, and single-use delivery links.

Analyze — runs, audit, live log

Every run keeps its steps, tokens and cost. Watch a live trace, read the audit trail, filter the live event feed (all · chat · tools · brain steps · errors) and export runs as JSONL for training.

Insights

Tasks, brain calls, tokens and cost — per day, provider, model and agent. Brain modes are compared from real runs: auto · cheap · balanced · deep · manual.

Agents

Per-agent model and system prompt, capabilities, skills and tool toggles (20 built-in tools), shell allow-list, allowed file paths, per-agent chat, start / stop / update.

Devices

Live telemetry: CPU, memory, disk, uptime, load, cores, version and last seen. One Remote Agent key connects a machine, and device tokens can be revoked.

Library and custom commands

Skills synced from ~/.remoteagent/skills, plus 232 commands across 9 categories: pick an allow-list or allow everything with *. Your own commands (dynamic tools) push to your devices.

Playground

Test any model with a stored key and see the response, provider, token usage and latency before an agent ever uses it.

Security in the console

Two-factor authentication (authenticator, backup codes, email code), approval requests you approve or deny, CSRF-protected writes, and every query scoped to your user.

What is not built yet

Cost is measured and reported per run. A user-set spending cap does not exist yet — the console says so on the settings page. Plans limit agents, runs and steps, not euros.

Chat

You talk. It works. You watch.

Every reply comes back with the steps behind it — what it thought, which tool it called, what came out, and what it cost.

console · chat with Assistant
Assistantrunning
MacBookonline
allowlist shellaudited
Workspaceread/write
Open Safari and go to YouTube's most played song
think — the user wants Safari, then YouTube… tool.call apps — launch Safari tool.result exitCode 0 — Safari is up tool.call browser — open youtube.com tool.result opened: true answer Done — Safari is open on YouTube verify checked the tool results
Message Assistant…Send

Example 1 from the published docs — a real captured run: 9,947 tokens · ≈ $0.0028 · deepseek-chat. The layout is the real chat hub.

Watch it think

The live trace streams while a run is going: think, tool.call, tool.result, answer, verify. If nothing is recorded yet, the console says exactly that instead of filling the space.

Stop it, correct it, rate it

Cancel a running task, thumbs-up or thumbs-down a reply, and pick which workspace an agent may touch before you send.

No agent? No setup screen

Your first message creates the helper automatically. If no device is online yet, the cloud answers inline instead of failing.

Workspace

Real files. Visible changes. Verified delivery.

Your files stay on your machine — the cloud only queues the operations your device runs.

workspace · My Workspace
FilesChangesPreviewRunsAgents
index.htmlstyles.cssapp.jsassets/

<!doctype html> <html lang="en"> <head> <meta charset="utf-8"> <title>My app</title> </head> <body> … </body> </html>

Acceptance

Renders, no console errors

Evidence

Verdict recorded against the run

content hash recorded on every change · versions kept

UI preview — the real workspace layout, with example content.

Operations run on your device

Read, write, patch, rename, copy, trash, restore, diff. The cloud queues typed operations; the device executes them with a lease and reports the result hash back.

Changes you can read like a diff

Every save becomes a version with a before → after hash, so you can see what moved instead of trusting a summary.

Preview without exposing your machine

Start and stop a preview server, opened in a locked-down frame — default-src none, nothing else reachable from it.

Acceptance and evidence

Write the criteria, let a run produce the evidence, and keep the verdict with the workspace. When a device mutates a file, the run is verified against the recorded hash and conflicts are reported.

Hand it over safely

Grant an agent review-only or read/write access, and mint a single-use delivery link when the work is done.

Examples

Real runs. Real cost.

Traces published in the docs, captured by the control plane and anonymized — nothing re-drawn.

Google Maps clone
one prompt24 steps76.5k tokens≈ $0.04
App Store clone
one prompt30 steps48.2k tokens≈ $0.02
Chess game
one prompt9 steps24.2k tokens≈ $0.0079

And 232 commands across 9 categories — the allow-list you pick from, or allow everything with *.

files 47system 47dev 39util 25network 23process 18packages 14media 11power 8
Skills, not prompts

Six built-in skills ship with the agent: briefing, disk-health, network-health, service-health, web-research and certificate-expiry. A skill is a written procedure the agent follows, and you can switch each one on or off per agent.

Your own commands

Add a dynamic tool (name, description, command, parameters) and it is pushed to your devices automatically — no redeploy.

Export the runs

Download finished runs as JSONL with steps, tokens, cost and your thumbs-up or thumbs-down — training data from your own work.

For you

Pick the life you actually live.

Same agent, different jobs. Every card names the real tools or the real skill behind it.

what people ask it to do
Run the tests and tell me what broke

It runs your dev commands, reads the failures, and explains them in order of importance.

shellgit · node · npm · pnpmfiles
Ship a change without surprises

Edits land in a workspace with a diff, a content hash and a version you can roll back to.

workspacediff + versionsacceptance + evidence
Tidy a messy folder

Searches, renames, copies — and moves deletions to a trash you can restore from.

files searchrename · copytrash + restore
Morning briefing

One message: how the machine is doing, the top headlines, and anything you asked it to remember.

briefing skillsysinfomemory
Which disk is about to fill up?

Checks every volume, flags the ones over 85%, and proposes one safe cleanup — it never deletes on its own.

disk-health skilldf -hreport only
Open things on my machine

Launch apps, open a page, speak a line out loud, or convert a photo — the everyday commands, on your side.

apps · browseropen · say · osascriptnotify
Research it properly

Two or three different searches, the best pages fetched, disagreements called out, a note saved for next time.

web-research skillwebmemory
Explain this codebase to me

Walks the tree, reads the files, and answers in plain language. Ask again tomorrow — it remembers.

files tree · readmemoryrecall
Revise with me

Reads your own notes and documents, then quizzes you on them. Your material never leaves your machine.

files readmemorystays local
Watch my services

Collects status and safe recent logs, proposes a restart, and waits for your OK. It does not restart on its own.

service-health skillprocesswaiting for your OK
Network down? Diagnose it

Runs read-only checks, classifies the fault, prepares a plan — and changes nothing without approval.

network-health skillip · ping · traceroutediagnose only
Certificates expiring soon

Reads validity dates, flags what is near expiry, prepares the renewal. Private keys are never read or shown.

certificate-expiry skillmetadata onlyno private keys

Skill names and commands come from the real catalogue. The runbooks diagnose and ask; they do not act alone.

Install — a real app, not a website

Runs on the computer you already own.

Install it once. After that you just type what you want — in the browser console, or in the app on your machine.

WINWindows 10 & 11
irm https://remoteagent.online/install.ps1 | iex

Run it in PowerShell. It installs the remote-agent command onto your PATH.

MACmacOS
curl -fsSL https://remoteagent.online/install.sh | bash

Apple Silicon and Intel. Then remote-agent login once.

NIXLinux & WSL2
curl -fsSL https://remoteagent.online/install.sh | bash

Same installer. Works behind NAT — there are no ports to open.

PIRaspberry Pi
curl -fsSL https://remoteagent.online/install.sh | bash

ARM boards included. Leave it on as your always-on helper.

Node.js 20+MIT licenceNo inbound portsWorks behind NATOne key per deviceRevoke anytime
remote-agent · the local app
  1. Install — one command for your operating system.
  2. Login — paste your Remote Agent key once. The key is shown once, and you can revoke it any time.
  3. Use it — in the browser console, or in the local app: remote-agent gui.

Your device dials out to the cloud. Nothing connects in, so nothing needs opening on your router.

Platforms, licence and requirements are the real published ones.

remoteagent.online × llminone.com

Create your dream in one click.
Keep it alive in a second.

Two open-source projects, two halves of one wish: one makes it real, the other keeps it alive. The thing you imagine in the browser does not die as a demo — it gets a machine, a budget and a memory.

① Create — llminone.com

Dream Machine

One sentence becomes a working thing, live in your browser.

  • A self-contained app, game or tool from a prompt
  • Built by DeepSeek, streamed while it is made
  • Refine in place, add pages, 30-step version timeline
  • Share by link, with owner / editor / viewer roles
  • 120 example prompts across 13 categories

llminone.com ↗

one button
the handoff

② Keep alive — remoteagent.online

Control plane

The build gets a home: real files, a keeper and a record.

  • Runs on your machine, on real files — not only a sandbox
  • Policy and a command allow-list before anything happens
  • Keys in an AES-256-GCM vault; tokens and cost per run
  • Acceptance criteria and evidence with a verdict
  • Content-hash versions, conflicts reported
runningguardedremembered

remoteagent.online ↗

① A wish becomes a buildOne sentence in Dream Machine — an app, a game, a tool, alive in the browser.
② The build gets a homeOne button. It becomes a workspace with real files and a content hash.
③ The home stays aliveAn agent runs it on your machine, under your policy, with the record kept.

It does not end when the demo ends. One vault. One budget. One record — every dream gets a keeper.

not integrated yet — the direction we are building toward