FAQ

Grouped by topic. If your question isn't here, dive into the relevant tutorial, or file a ticket via Settings → Extended Capabilities → Extended Features → Feedback Helper.

About the App

Is this an iPhone app or a Watch app?

Both. ETOS LLM Studio runs on both iPhone and Apple Watch, with sync between them. See Sync & Backup.

Do I need an Apple Watch to use it?

No. The Watch is a featured surface, but the iPhone end is fully complete on its own.

Does this only support OpenAI?

No. Four API formats are supported:

• OpenAI Compatible (Chat Completions) — covers official OpenAI and nearly every third-party relay
• OpenAI Responses — OpenAI's newer endpoint
• Anthropic — Claude
• Gemini — Google

If your service speaks one of these, you can connect. See Add Your First Provider.

Is the app paid?

No. The app is free with no subscription. You pay only the LLM provider you connect (OpenAI / Anthropic / Google, etc.).

Where does my data go?

By default, nowhere. Everything stays in the local SQLite database. Data leaves only if you explicitly:

• Enable iCloud sync (encrypted upload to your iCloud)
• Enable Apple Watch sync (LAN direct, no external server)
• Upload backups to your S3 / R2
• Send a conversation as a request to your connected LLM (intrinsic to using the model API)

See Sync & Backup.

Setup & Configuration

Where do most people get stuck first?

Provider configuration. The advice: connect only one stable model first, get text chat working, then enable image, voice, tools, and proxies. See Add Your First Provider.

Why do so many features feel buried?

ETOS packs everything into the Settings tab, keeping the main UI for chat. That keeps daily use clean but means Interface Tour is worth a read.

When you can't find something, use this rule:

• Execution (chat, attachments) → Chat tab
• Governance (providers, tools, memory, worldbook, sync) → Settings tab

Chat says "Select a model to start"

Go to Settings → Current Model → Model and explicitly pick one.

If the list is empty ("No models available. Please enable one under Providers & Models."), you added a provider but didn't enable a model row — open Providers & Models and toggle on the model you want.

My requests keep failing with "Authentication failed"

Usually an API key issue:

1. Re-copy the key from the vendor console
2. Make sure there's no leading/trailing whitespace
3. Don't prepend Bearer manually — the app adds it
4. Check if the key has expired

See Add Your First Provider → Common Errors.

Feature Choices

Memory vs Worldbook — how do I choose?

This info …	Goes in
Used repeatedly, like a knowledge block	Memory
Only useful when a specific keyword / scene appears	Worldbook

Simple examples:

• "Reply in Chinese by default" → Memory
• "When Character X shows up, recall their background" → Worldbook

See Memory & Worldbook.

What's Daily Pulse?

A daily auto-generated stack of cards on "what's worth looking at today," based on your local signals (recent chats, memory, feedback). See Daily Pulse and Daily Pulse Internals.

Are Tools and MCP required?

No. You can use ETOS purely as a chat client. But if you want richer workflows (let the AI search files, call external services, run a Shortcut), the tool system is well worth the time. See Tools & MCP.

Is memory cloud-hosted?

No. Embeddings call a cloud API (during vectorization), but the vector database itself is local SQLite.

Can different sessions use different models?

Yes. The top-right of the chat screen switches the current session's model. It only affects this session, not the default.

Multi-device / Sync

Do I have to enter API keys on the Watch?

No. Strongly recommended to do all configuration on iPhone and let sync carry it to the Watch. See Using Apple Watch.

Does iCloud sync include API keys?

Yes. When iCloud sync is on, all data (including API keys) is encrypted and uploaded to your iCloud. Apple can't see it, but make sure your Apple ID has two-factor authentication.

If you don't want keys on iCloud, use only Apple Watch sync (LAN direct, never uploads).

Will uninstalling lose my data?

Yes — irrecoverably. Before uninstalling, take a Full Snapshot (Settings → Display & Experience → Sync & Backup → Database Snapshot → Full Snapshot) and stash it on iCloud Drive or S3.

Debugging & Feedback

How do I debug effectively?

Identify which layer is broken:

Symptom	Likely layer
No model connects	Network / proxy
Some models don't connect	That provider's key / Base URL
Models connect but reply is off	System prompt / memory / worldbook interference
Tools can't be called	Approval policy / Tool Center master switch / worldbook isolation
Sync down	Both ends on same network? / iCloud status

Then jump to the relevant tutorial. See Debug & Feedback.

How do I file effective feedback?

Settings → Extended Capabilities → Extended Features → Feedback Helper → + New Ticket

Include:

• Exact reproduction steps (taps, fields filled)
• Environment info (auto-captured)
• App logs (redacted, the date folders are easy to find)
• Screenshots (optional)

Submit after the PoW completes. Status syncs automatically.

My ticket hasn't been answered

ETOS is open-source / spare-time pace, not 24×7 support. First response usually within 24–72 hours. If a week passes, follow up on the ticket or duplicate on GitHub Issues.

Other

Can I contribute code / file a PR?

Yes. Repo: github.com/Eric-Terminal/ETOS-LLM-Studio.

How do I contribute to this doc site?

Source lives in docs/site/. Local preview:

cd docs/site
pnpm install
pnpm docs:dev

Chinese version at the root, English version mirrored under en/. Submit a PR after edits.

Still have a question?

• Try Overview — maybe a step was skipped
• Try Interface Tour — maybe the feature is just hiding
• Try Hidden Gems — maybe a gesture was missed
• File a ticket / GitHub Issue