Your server control room, in your pocket
Cockpit AI
Server Monitor
Connect to your Linux servers over SSH, watch CPU, memory, disk and network in real time, and ask your own LLM about a stuck log in plain English. Runs on iPhone, iPad and Android.
At a glance
Check your servers
from your pocket.
No need to open a laptop and spin up an SSH window — see load, resources and service status for your registered servers right in the app. Nothing to install on the server side: it works over plain SSH, no agent required.
Real-time dashboard
CPU · memory · disk · network · load average · uptime · service status as cards on one screen. Reorder the widgets to your liking.
Multiple servers
Register several servers and switch between them with a tap. SSH credentials live in the OS secure store (Keychain / EncryptedSharedPreferences).
Metric history
See recent resource trends as graphs. "Did memory spike overnight?" — answered from your phone.
Threshold alerts
Push notifications when CPU, disk and more cross a line you set. Rules are defined per server.
Command palette
Save the commands you run often and fire them in one tap. Restarts, disk cleanup and other routines — like buttons.
SFTP backup & restore
Back up your settings and server list to your own server over SFTP, and restore on a new device from the dashboard shortcut.
New in v1.2 — AI Log Diagnostics
Ask a stuck log
in plain English.
Instead of copying an error log into a search engine, send your server logs straight to an AI and ask "why am I getting a 502?" in plain language. The key part is BYOK (Bring Your Own Key) — it runs only on an LLM you choose, and our servers are never part of the flow.
Your LLM, as-is
Point it at a local Ollama / LM Studio, or paste an OpenAI · Gemini · Anthropic API key. Anything that speaks the OpenAI-compatible endpoint will connect.
Per-service presets
Seven presets: nginx · Apache · MySQL · PostgreSQL · Redis · Docker · system. Pick the log source and start diagnosing right away.
Two-step shell confirm
Any shell command the AI suggests goes through a two-step confirmation before it runs. Dangerous patterns like rm -rf /, mkfs and dd are blocked outright.
Tuning guide
Config changes the AI proposes aren't applied blindly — you're shown what and why first, then you decide.
💡 With a local LLM there's no token cost and nothing leaves your network
Run Ollama or LM Studio on the same network and you can analyze logs for free, with no data leaving your machines. Prefer a cloud model? Just paste your own key. Either way, the key and endpoint stay in your phone's secure store.
Security — where data flows
Logs go from your phone
to your server, and nowhere else.
Cockpit doesn't own a server to receive your server data. Both the SSH connection and the AI analysis call go directly from your phone to a target you choose. We're never in between.
SSH is phone → your server, directServer credentials and sessions never pass through a developer server. Connection details are stored only in the OS secure store (iOS Keychain / Android EncryptedSharedPreferences).
AI log diagnostics is BYOK — straight to your endpointLogs and prompts are sent directly to the LLM endpoint you configured (local Ollama / LM Studio, or your own API key). They never relay through our servers, and with AI off there's no call at all.
Risky shell commands: two-step confirm + pattern blockCommands the AI proposes go through a separate confirmation step, and system-destroying patterns (rm -rf /, mkfs, dd, …) are blocked automatically. You always make the final call.
No sign-up, no loginNo account to create. The app talks to nothing but your server and your LLM — no developer backend.
Free tier shows a banner ad (Google AdMob)Free users see a bottom banner ad; on iOS you'll be asked for tracking consent (ATT). Upgrade to Pro and the ads are gone.
No developer server collecting logs or credentialsThere's no backend gathering your metrics, logs or SSH keys. You hold all the control.
Data flow — drawn out
⚠️ User responsibility
Cockpit is a tool for servers you run yourself. Even when a command is AI-suggested, the final responsibility to run it is yours — review it and approve. Usage and billing for any third-party API key you enter via BYOK is managed in that provider's own console.
Features
Full feature list
The core monitoring features are free. Ad removal and AI diagnostics unlock with Pro. AI log diagnostics runs on your own LLM key (BYOK) — and your first 30 days unlock every Pro feature for free.
- SSH multi-server connect & manageFree
- Real-time dashboard (CPU·memory·disk·network·load·uptime)Free
- Service status viewFree
- Metric history graphsFree
- Threshold-based push alert rulesFree
- Command palette — save & run frequent commandsFree
- SFTP settings backup & restoreFree
- Korean / English UIFree
- Ad removalPro
- AI log diagnostics — analyze logs in plain EnglishPro · BYOK
- Per-service diagnostic presets (nginx·Apache·MySQL·PostgreSQL·Redis·Docker·system)Pro · BYOK
- AI shell command suggestions — 2-step confirm · risky patterns blockedPro · BYOK
- AI tuning guidePro · BYOK
Pricing
30-day free trial,
then you decide.
Core monitoring stays usable for free. Pro (ad removal + AI diagnostics) is free for the first 30 days from install. When the trial ends, only the Pro features lock — the free features keep working.
First 30 days — free
All Pro features + all free features — no sign-up.
- All free monitoring features
- Ad removal
- AI log diagnostics (BYOK)
- Per-service diagnostic presets (BYOK)
- AI shell suggestions & tuning guide (BYOK)
※ After 30 days the free monitoring features stay; only Pro features lock. Subscriptions are in-app purchases and can be cancelled anytime from your Apple ID / Google account settings.
Supported
On iPhone, on Android.
A single Flutter codebase runs natively on both platforms. On the server side, nothing beyond standard SSH is required.
FAQ
Frequently asked
Do I need to install an agent on my server?
No. It works over plain SSH — there's no monitoring daemon or extra package to install. If you already SSH into a server, just register it and you're good to go.
If I use AI log diagnostics, where do my logs go?
It's BYOK (Bring Your Own Key). Logs go only to the LLM you choose — use a local Ollama / LM Studio on your network and nothing leaves it; paste an OpenAI / Gemini / Anthropic key and it calls that provider's API directly. Either way our developer servers are never in the flow, and with AI off there's no call at all.
What if an AI-suggested command breaks my server?
Shell commands the AI proposes go through a two-step confirmation, and you always make the final call. System-destroying patterns like rm -rf /, mkfs and dd are blocked, so no run button even appears. Still, since it's your own server, review the command and approve it yourself.
How far can I get for free?
SSH multi-server connect, the real-time dashboard, metric history, threshold alerts, the command palette and SFTP backup/restore are all free. The free version shows a bottom banner ad. Upgrade to Pro to remove ads and unlock AI log diagnostics — and your first 30 days unlock all of Pro for free.
Are my SSH keys and passwords stored safely?
Credentials are stored only in the OS secure store (iOS Keychain / Android EncryptedSharedPreferences). Other apps and other users can't reach them, and they're never sent to a developer server.
Can I use the AI features without a local LLM?
Yes. Paste an API key for a cloud model like OpenAI, Gemini or Anthropic and you're set. To avoid cost, run Ollama / LM Studio on a PC or NAS and connect to it — both expose an OpenAI-compatible endpoint.