Skip to content

FAQ

No, not necessarily. For Anthropic you can sign in with your Claude account through OAuth, in the desktop onboarding or with wayland-core --login, and use your subscription with no key to paste. Other providers use an API key, which you add in Settings -> Models (desktop) or your config file (engine).

Yes. Wayland runs on your machine. Conversations, memory, and config live locally. Model providers see only the messages you send them, the same as any direct API use. If you turn on sync, backups are end-to-end encrypted with a passphrase only you hold.

What is the difference between the desktop app and Wayland Core?

Section titled “What is the difference between the desktop app and Wayland Core?”

The desktop app is the full command center: assistants, teams, channels, scheduling, voice, image generation, memory, and remote access behind a UI. Wayland Core is the engine underneath, available on its own as a command-line agent for scripts, CI, and headless hosts. They share config and run the same loop. See What is Wayland.

Do I have to install Wayland Core separately?

Section titled “Do I have to install Wayland Core separately?”

Only if you want it on your PATH. The desktop app bundles a matching wayland-core binary and prepares it on first run, so a clean desktop install works immediately once you add a provider. Install the standalone CLI when you also want to script the engine. See Install Wayland Core.

How many assistants, teams, and channels ship?

Section titled “How many assistants, teams, and channels ship?”

The desktop app defines 29 built-in assistant presets and loads additional assistants from a bundled extension; that bundle currently contributes 55 more entries (27 team launchers and 28 domain specialists), for a combined starting catalog of around 84 assistants before any extras you add. Team launchers include both Standing Companies (persistent virtual orgs with dedicated teammates and rituals) and ad-hoc team configurations; the split between those two forms reflects the current bundle version. The workflow library contains 70 bundled workflows. Channels span 25 plugin directories across three tiers. The provider catalog covers 32 providers.

Eight built in: Read, Write, Edit, Bash, Grep, Glob, Delegate, and ToolSearch. MCP servers add more on top. See Architecture Overview.

Three levers help. Compaction keeps long sessions within budget automatically (three tiers, plus --compaction for output). Prompt caching reuses stable context for large savings on repeated runs. And you can assign cheaper models to high-volume assistants while keeping a strong model where it matters. See Configuration.

Run wayland-core --config-path to print the global config location for your OS. A project can override it with a .wayland-core.toml in its directory. See Configuration.

The desktop app updates itself through its built-in updater. For the engine, run wayland-core self-update, or wayland-core self-update --check-only to see what is available first. Updates are signature-verified before they apply.

Tools keep asking for approval. Can I stop that?

Section titled “Tools keep asking for approval. Can I stop that?”

Yes, deliberately. Approve and remember a specific call to stop being asked for that kind. For unattended scripts, --auto-approve skips the prompt and --force removes the gate entirely. Use --force only for trusted, scripted runs. See Single-shot and REPL.

Something is broken. Where do I report it?

Section titled “Something is broken. Where do I report it?”

Bugs go to GitHub Issues. Questions and ideas go to GitHub Discussions, and the community gathers on Discord. The community is not a paid support desk: Wayland is free and in beta, so please be specific and patient.