ETOS LLM Studio Docs

English

简体中文

English

简体中文

Theme

ETOS LLM StudioA Native AI Client for iPhone + Apple Watch

Bring your own API key. Local-first storage. Two-way device sync. Zero middle server. Talk to OpenAI / Claude / Gemini / any OpenAI-compatible service — with memory, worldbook, MCP, Skills, Daily Pulse, and Shortcut tools.

First chat in 10 minutes
Module deep dives
GitHub
ETOS LLM Studio
📱

Native on iPhone + Apple Watch

Both devices have their own interaction paths, reorganized for each screen size — not a phone app awkwardly shrunk onto a watch.

🔌

Your Key, Your Data

Model requests go straight from the device to the provider — no middle server. Conversations live in a local SQLite database; export an ETOS bundle any time.

🧠

Long-term Memory + Worldbook + Daily Pulse

More than one-shot chat. Cross-session facts, keyword-triggered knowledge patches, scheduled proactive briefings — all wired into the conversation flow.

🧰

MCP / Skills / Shortcuts / File Tools

Give the model "plugin" capabilities. MCP for external services, Skills for local capability bundles, Shortcuts to let the model trigger iOS automations.

🎛️

Advanced Request Configuration

Multi-key rotation, custom headers, parameter expressions, raw JSON body, per-provider and global proxies — whatever weird "compatible" endpoint exists, you can probably hit it.

🔄

Cross-device Sync + Third-party Import

Direct LAN sync between iPhone and Watch. Migrate from Cherry Studio, RikkaHub, Kelivo, or the official ChatGPT export.

What This Doc Site Does ​

ETOS LLM Studio hides every feature inside Settings and keeps the main UI for chat. That keeps daily use clean, but it also means lots of features are easy to miss even after you install the app.

This documentation has three jobs:

  1. Step-by-step tutorials — from install to first reply, with exact taps, fields, and "you succeeded when you see X."
  2. Feature deep dives — every module is introduced as what it is, what problem it solves, what happens without it, then the configuration.
  3. Design rationale — if you want to know why ETOS works the way it does (context assembly, Daily Pulse signal sources, Worldbook injection rules), the design docs cover it.

Where to Start ​

You're …Start here
New, haven't downloadedOverview (10-minute path)
Installed but stuckAdd Your First Provider
Chatting; want a full tourInterface Tour
Going deep on one featureModules
Curious about the designDesign
Hunting for tricksHidden Gems
Stuck on a problemFAQ

Screen Previews ​

Many features don't announce themselves. This doc site fills in their physical locations, recommended use order, and the explanations the app deliberately keeps off-screen.

Chat screen
The chat screen isn't just a message list. Model switching, attachments, tool toggles, thinking, TTS read-aloud, and message export all live on one screen.
Settings screen
Settings is the cockpit. Providers, tools, memory, sync, Daily Pulse, and the feedback system are all grouped here.