Doc View Terminal - Writer Tools

Overview

Doc View is an AI writing terminal — a command-line interface where you type a writing command and the AI generates long-form content in return. It is purpose-built for producing structured documents: blog posts, essays, articles, short stories, and presentation decks.

The screen is divided into two panes. The left pane is the terminal — you type commands here and see AI generation progress. The right pane is the live preview — rendered markdown or structured presentation slides appear here as generation completes. A split toggle in the header collapses the left pane to a thin strip if you want more reading space.

Terminal Workflow

Doc View is intentionally keyboard-first. Type a command, press Enter, and the AI writes. You can chain multiple commands in sequence to build a content package — blog post + matching social captions + a presentation version — all in one session.

Writer Commands

Doc View recognises these top-level writing commands:

Command	Output	Typical Length
write-blog [topic]	Blog post with introduction, body sections, and conclusion	600–1200 words
write-essay [topic]	Academic-style essay with thesis, arguments, and references section	800–1500 words
write-article [topic]	Journalistic article with headline, lead paragraph, and sections	500–900 words
write-story [topic]	Short narrative story with characters, conflict, and resolution	800–2000 words
write-presentation [topic]	Structured slide deck with title, content, and conclusion slides	8–15 slides

Commands accept a plain language topic description. You can be as brief as "write-blog remote work" or as detailed as "write-blog the productivity benefits of asynchronous communication in distributed engineering teams, targeting a technical audience".

Topic Specificity

Specific topics produce focused, actionable content. Vague topics produce generic content. "write-essay leadership" produces a generic essay; "write-essay why psychological safety is the foundation of high-performing engineering teams" produces something that could actually be published.

Output Modes

Doc View supports two output modes, selectable in Settings → Terminal → Block Behaviour → Output Display:

Modal (default) — a summary row appears in the terminal when generation completes. Click the View button or wait for the modal to auto-open (controlled by the Auto Open Modal toggle in settings). The full document opens in a full-screen reader with clean typography and no terminal chrome around it.
Terminal — the rendered markdown content appears inline in the terminal block, immediately below the command. A View button is still available to open the full-screen reader. This mode is better if you want to review output without leaving the terminal workflow.

The auto-open behaviour applies only in Modal mode. In Terminal mode, content is always shown inline immediately.

Viewing and Editing

Click the View button on any generated document to open it in the full-screen Document Reader. The reader has two modes:

Read Mode

Clean full-screen view of the rendered document. Typography is optimised for long reading sessions. Use the Download button to save as Markdown, or the Copy button to copy the full text to the clipboard.

Edit Mode

Click the Edit button in the reader toolbar to switch to edit mode. The title becomes a text input and the body becomes a resizable textarea with monospace font. Edit freely and click Save to persist changes. The updated version is saved to My Reports and reflected in the terminal summary row.

AI + Human Editing

The recommended workflow is: use the AI to produce a first draft at full speed, then switch to Edit mode to refine tone, add specific data points, and adjust structure. This is typically 5–10× faster than writing from scratch.

Writer Engine Settings

Doc View supports three AI backends, selectable in Settings → Doc View → Writer Engine:

API (default) — uses the configured API. Sub-mode determines which API: Simplified.com API for blog and article multi-stage generation; Cohere API for storyteller; Anthropic API (claude-sonnet-4-6) for everything else.
Local CLI — calls the Claude CLI (claude --print) on the server. Falls back automatically when ANTHROPIC_API_KEY is not set. Good for offline or privacy-sensitive environments.
Local LLM — uses BERT for intent classification only. Text generation commands are not supported in this mode — use it only when you need to classify commands without generating output.

Export

All documents generated in Doc View can be exported from the full-screen Document Reader:

Download as Markdown — saves a .md file to your device. Preserves all headings, formatting, and structure.
Copy to Clipboard — copies the full plain text (with markdown syntax) to the clipboard for pasting into other editors, email, or CMS platforms.
Save to My Reports — persists the document to your My Reports library (visible in the PM module), where it can be retrieved, edited, and shared at any time.

Frequently Asked Questions

What's the difference between Modal and Terminal output mode?

Modal mode (the default) keeps the terminal focused on commands — each generation completes as a compact summary row, and the full document opens in a dedicated full-screen reader (automatically, if the Auto Open Modal setting is on). Terminal mode instead renders the full markdown content directly inline in the terminal block below the command, which is useful if you want to scan multiple generated documents without leaving the command flow. Both modes still offer a View button to open the full-screen reader on demand. Switch between them anytime in Settings → Terminal → Block Behaviour → Output Display.

Can I edit content after the AI generates it?

Yes. Open any generated document with the View button, then click Edit in the reader toolbar. The title becomes an editable text field and the body becomes a resizable monospace textarea. Click Save to persist your changes — the updated version is written back to My Reports and the terminal's summary row reflects the edit immediately. The recommended workflow is to let the AI produce a fast first draft, then use Edit mode to refine tone, insert specific data, or restructure sections.

Which writer commands are available, and how specific should my topic be?

Doc View supports five commands: write-blog, write-essay, write-article, write-story, and write-presentation, each producing a different document shape and typical length (see the Writer Commands table above). Topic wording matters — a vague prompt like "write-essay leadership" produces generic filler, while a specific prompt like "write-essay why psychological safety is the foundation of high-performing engineering teams" produces something close to publishable. Longer, more specific topic descriptions consistently produce more focused, actionable output.

What happens if I switch the Writer Engine to Local LLM?

Local LLM mode uses BERT purely for intent classification — it cannot generate long-form text, so all write-blog/write-essay/write-article/write-story/write-presentation commands will fail in this mode. Use Local LLM only if you need lightweight command classification without generation. For actual writing, use API mode (the default, routing to Simplified.com, Cohere, or Anthropic depending on the command) or Local CLI mode, which calls the Claude CLI on the server and is a good option for offline or privacy-sensitive environments.