Terminal Settings

Configure the Writer Engine, AI backend, block display behaviour, and output modes for all terminal-based tools in TotalApp.

Writer Engine

The Writer Engine controls which AI backend is used when terminal commands generate text — blog posts, essays, articles, stories, and other long-form content. Three engines are available:

Engine	How it works	Best for
API (default)	Calls a hosted AI API (Anthropic Claude or Cohere, depending on API Mode). Requires an internet connection and a valid API key configured on the server.	Best quality; recommended for production use.
Local CLI	Runs the `claude --print` command locally. Falls back automatically when `ANTHROPIC_API_KEY` is absent on the server.	Offline or air-gapped environments with Claude CLI installed.
Local LLM	Uses a locally running BERT model for intent classification only. Text generation commands are not supported and will return an error.	Classification-only tasks; not for writing.

API Mode (sub-selector)

When the Writer Engine is set to API, a secondary selector appears to choose the API provider:

Mode	Provider	Used for
Simplified	Simplified.com API (`/text-to-article`)	Blog (multi-stage), Article, Ad copy, Scenario generation.
Cohere	Cohere API (`command-r7b-12-2024`)	Storyteller. Other commands fall back to the Claude endpoint.

Block Behaviour

These settings control how generated content blocks appear inside the Doc View terminal after an AI writer command completes.

Option	Values	Description
Output display	Modal / Terminal inline	Modal: a summary row appears in the terminal; the full content opens in a pop-up viewer. Terminal inline: content renders directly inside the terminal block.
Auto open modal	On / Off	When Output display is set to Modal, automatically opens the viewer as soon as generation completes. Only visible when Modal mode is active.
Block action buttons	Icons / Icon+Text / Text	Controls how the View, Edit, Export, and Delete buttons appear on writer blocks: icon-only, icon with label, or text-only label.

Tip

Use Modal mode when running multiple write commands in one session — the terminal stays clean and each document opens separately. Switch to Terminal inline for a quick read without leaving the terminal context.

Terminal Enable / Disable

Individual terminals (Terminal Editor, Dev Hub, Security Hub, Graph View, Canvas Terminal) can be enabled or disabled here. Disabled terminals are hidden from the sidebar and cannot be opened. This is useful for teams who want to expose only the terminals relevant to their workflow.

FAQ

What happens if I select API mode but no API key is configured?

The server falls back automatically to the Local CLI engine if ANTHROPIC_API_KEY is absent. If Claude CLI is also not installed, write commands will return an error prompting you to configure an API key.

Can I switch engines mid-session?

Yes. Engine changes take effect immediately on the next command. Blocks already generated in the current session are not affected.

Does Local LLM support any generation commands?

No. Local LLM uses BERT for intent classification only. Attempting a generation command (blog, essay, story, etc.) in Local LLM mode returns an explicit error message.