Onepilot
Show HN: An agentic control pane for iPhone
Last verified:
What is Onepilot?
Onepilot is a mobile-first developer IDE for iPhone that combines a full SSH client, remote file browser, git tooling, cron management, and built-in AI agent deployment to let developers manage servers and ship code from their phone. The app deploys and runs terminal-native AI agents (OpenClaw, Hermes, Claude Code, Codex, with Paperclip coming) on any host reachable over SSH and ties each agent to one of 25 supported LLM providers and messaging channels like Telegram, Discord, or Slack. Key features include a one-tap deploy wizard for installers, a live command terminal with full VT100 emulation and touch-optimized input, a file browser with syntax highlighting for 20+ languages, git diff/commit/branch views, cron scheduling and management, and a central command center that shows live connection status, running agents, and scheduled jobs. Onepilot stores all credentials in the iOS Keychain and forwards API keys directly to the target host (no traffic routed via Onepilot servers), making it suitable for developers who need remote server control, mobile DevOps, and on-device secret management while using autonomous coding agents.
Onepilot pricing
Pricing model: Freemium
Free to get started: Onepilot is free and includes unlimited SSH connections, unlimited agent deployments, the full terminal, file browser, git tooling, cron job management, Soul Designer, 25 LLM providers support, messaging channel integrations, and no hosting fees; users bring their own servers and LLM API keys. There are no advertised paid plans or hosting charges on the website; the business model expects customers to use their own infrastructure and provider API keys. No trial, subscription, or tiered feature list is shown—the site emphasizes that everything needed to connect, deploy, and manage is included from day one, with credentials stored locally in the iOS Keychain.
Onepilot pros
- Deploys AI agents directly onto any SSH-reachable host
- Supports 25 LLM providers out of the box
- Built-in one-tap deploy wizard for OpenClaw, Hermes, Paperclip path
- Full interactive VT100 terminal optimized for touch
- Remote file browser with syntax highlighting for 20+ languages
- Git integration with diffs, commits, and branch workflows
- Cron job creation and scheduled automation from the phone
- Central command center showing live status of servers and agents
- Stores SSH keys and LLM API keys in iOS Keychain (Secure Enclave)
- Allows swapping models, rotating keys, and editing system prompts on the fly
- Can bind agents to messaging channels (Telegram, Discord, Slack)
- No hosting fees—you use your own infrastructure and API keys
- Unlimited SSH connections and unlimited agent deployments
- Deploys Claude Code and OpenAI Codex as terminal-native agents
- Brings agent management and terminal into a single mobile IDE
Onepilot cons
- iPhone-only native app (no first-class Android client documented)
- Requires an SSH-reachable server and user-managed hosting
- Users must supply their own LLM API keys (no hosted inference)
- Some agent frameworks (Paperclip) are not yet fully shipped
- On-device secret forwarding means setup friction for beginners
- Not a turnkey hosted agent platform—relies on user infra health
- No indication of built-in paid tiers or hosted compute options
- Advanced features (agent tuning, large-scale orchestration) require server knowledge
Frequently asked questions about Onepilot
What is Onepilot?
Onepilot is a mobile-first agentic IDE for iPhone that combines an SSH terminal, a file browser with syntax highlighting, git tooling, cron management, and an AI agent deployer so developers can manage servers and run coding agents directly from their phone.
Which AI agents can I deploy with Onepilot?
Onepilot ships deploy support for OpenClaw and Hermes, runs Claude Code and OpenAI Codex as terminal-native agents, and lists Paperclip on the roadmap; the deploy wizard installs the chosen framework on any host you SSH into.
What servers can I connect to?
Any host reachable over SSH: common targets include AWS EC2, Hetzner, DigitalOcean, Google Cloud VMs, Linode, Vultr, Oracle Cloud, Raspberry Pi devices, macOS and Linux machines, Docker containers exposing sshd, and corporate jump hosts.
How does Onepilot handle my credentials and API keys?
Onepilot stores SSH credentials and LLM API keys exclusively in the iOS Keychain (protected by the Secure Enclave) and forwards API keys directly to the target server during agent deployment; it does not route SSH traffic or credentials through Onepilot servers.
Which LLM providers are supported?
As of 2026 Onepilot supports 25 LLM providers, including Anthropic Claude, OpenAI, Google Gemini, Mistral, Groq, Perplexity, Ollama (for local models), OpenRouter, Together, and others, plus custom OpenAI-compatible endpoints.
Do I need to pay for Onepilot to deploy agents?
No—Onepilot’s website states it is free to start and includes unlimited SSH connections and unlimited agent deployments; however, you must provide and pay for any LLM API usage and your own server hosting separately.
Can I control agents from other apps or channels?
Yes—deployed agents can be bound to messaging channels (Telegram, Discord, Slack) so you can send tasks and receive results through those platforms as configured during deployment.
Is Onepilot an SSH client or a full IDE?
Onepilot is presented as a full mobile IDE: it includes a real interactive terminal as one part of an environment that also offers file browsing with syntax highlighting, git integration, cron management, and AI agent deployment and supervision.
How do I deploy an agent using Onepilot?
Deployment is a four-step wizard: add the target server with SSH credentials, tap Deploy Agent and choose a framework, pick an LLM provider and paste its API key (stored in Keychain), choose a messaging channel, and confirm—the wizard provisions the agent over SSH and reports when live.
Can Onepilot run on self-hosted or local models?
Yes—Onepilot supports local model integrations such as Ollama and allows custom OpenAI-compatible endpoints, so self-hosted or private models with standard APIs can be used as LLM providers for deployed agents.