Skip to main content
Aether’s settings screen is your control center for every aspect of the app’s behavior — from which LLM provider powers your chats, to how the agent thinks, to which extensions are loaded. Everything you need to tailor Aether to your workflow lives in one place.

Accessing Settings

Open the conversation drawer from the menu button, then tap Settings. Settings are organized into clearly labeled sections so you can find what you need quickly.

Settings Sections

Model Providers

Add, edit, and remove your LLM provider credentials here. Each provider entry stores an API key, OAuth credential, or environment configuration required to connect Aether to that service. For a full walkthrough of adding and managing providers, see the Providers page.

Default Models

Choose the default models used for new chat sessions, conversation titles, automatic naming, and compaction. Individual chats can override the chat default directly from the top bar’s model selector, so your global setting is just the starting point.

Agent Mode

Authorize Agent Mode with Shizuku or experimental Root access. Agent Mode adds isolated virtual-display tools; Aether’s tool-calling loop, shell commands, file tools, Skills, and MCP servers work without it.

Extensions

Browse every extension currently installed in Aether, install new extensions from a .zip archive or the npm-backed extension catalog, and review any extension errors logged at load time. You can also manage Native Mod Safe Mode from this section — if a compiled Kotlin/DEX mod is causing crashes, Safe Mode lets you disable it without losing your other extensions.

Agent Skills / MCP Servers

Manage installed Agent Skills and configure MCP (Model Context Protocol) server connections on their separate settings pages. Skills provide reusable instructions and resources. MCP servers can be selected per chat.

Scheduled Tasks

Set up recurring agent workflows that run automatically at specified times without any manual input. Configure each task’s prompt and interval, daily, or weekly schedule. For full details, see the Scheduled Tasks page.
Changes to provider credentials and model selection take effect immediately — no restart required.