Experimental Features

Settings > Experimental holds toggles for features that are still maturing. “Experimental” doesn’t mean unsafe — these are shipped and tested — but the surface area, defaults, or persistence shape may shift between releases. Toggling one off is always reversible.

Six toggles live there today.

Alternative Claude Code backends

Exposes Ollama (local), LM Studio (local), the OpenAI API, and OpenAI Codex (subscription) in addition to the default Claude Code path. Custom Anthropic / OpenAI endpoint kinds exist in the data model but the GUI doesn’t yet have an “Add backend” flow.

When this is off, every workspace runs the official claude CLI authenticated against Anthropic. When on, the chat header gets a provider picker and Settings > Models grows configuration sections for each provider.

→ Full setup and capability matrix: Alternative Providers

Claudette Terminal

Adds a read-only terminal tab that mirrors the agent’s shell commands and any background-task output as the agent runs. Useful for watching what an agent is actually doing in real time without scrubbing through the chat transcript, and for catching long-running shell processes the agent forgot to wait on.

Background-task completion handling stays enabled even when this toggle is off — the toggle only controls whether the dedicated tab shows up in the terminal panel.

Plugin Management

Surfaces the Claude Code plugin marketplace inside Claudette: install, enable, disable, update, and remove Claude Code plugins from a settings panel rather than the CLI. Adds the /plugin and /marketplace slash commands to the picker (these are reserved when the toggle is on, so user/project commands of the same name yield).

This is distinct from Claudette’s own Lua plugin system (SCM providers, env-providers, etc.), which is always available.

→ Where the commands live: Slash Commands

Claude Remote Control

Bridges your local Claudette workspace to a Claude Code session running on claude.ai (web or mobile). Drive an agent from your phone while looking at the diff on a laptop, or hand off a session between devices.

Has its own setup flow, troubleshooting steps, and constraints (some Claudette features are deferred while a bridge is live).

→ Full doc: Claude Remote Control

Community Registry

Enables the Plugins / Themes / Grammars community catalog inside Claudette. Browse third-party content, verify signatures, and install with one click. Off by default because it broadens the trust surface — you’re running code from people who aren’t you or Anthropic.

→ Trust model and signature verification: Community Registry & Trust

Usage Insights

Gates the Settings > Usage panel that shows Claude Code subscription details (session limits, weekly limits, extra usage) by querying Anthropic’s usage API directly. Requires a Pro or Max plan with standard login (env-var auth tokens don’t have the OAuth scopes the API needs).

When this is off, Claudette never reads your Claude Code OAuth tokens for usage data — they stay strictly in the keychain / .credentials.json for the claude CLI’s own use. The Claude Code sign-in flow in Settings > General remains available because agent credential failures need a recovery path independent of usage tracking.

→ What the panel shows and how it’s cached: Usage & Metrics

Stability outlook

Toggle	What’s still moving
Alternative Providers	Per-provider capability surfaces; gateway translation edge cases on long streamed responses.
Claudette Terminal	Layout and persistence story for the read-only tab.
Plugin Management	Marketplace UI; settings persistence for per-plugin config.
Claude Remote Control	Reconnect / device-handoff UX; deferred-feature list.
Community Registry	Content moderation; revocation flow; theme runtime loading.
Usage Insights	Cache invalidation around tier changes; richer breakdown views.

If a toggle moves to “default-on” or graduates out of experimental, the upgrade path leaves your existing settings in place — toggles are independent, never bundled.