Core Workflows

Scritorio should organize the author experience around clear, repeatable manuscript-development loops.

Open A Local Project

The author opens a local folder. Scritorio reads Markdown files, front matter, project settings, reports, assets, and index data from that folder. The project remains usable outside Scritorio. The same project should be operable from the desktop app and the CLI. For series workspaces, the author should be able to choose a current book while still accessing shared series context.

Write And Review

The basic writing loop:
  1. Open a scene, chapter, section, beat sheet, or note.
  2. Edit the Markdown source.
  3. Choose an editorial action.
  4. Preview the context that will be sent.
  5. Run the review.
  6. Save the report locally.
  7. Revise the manuscript.
  8. Compare versions or reports.
In a series, reviews should make context scope explicit: current book only, current book plus selected series notes, or selected prior books.

Plan And Draft

Authors should be able to plan before drafting without being forced into one methodology. A beat-driven workflow may look like:
  1. Create or open a manuscript unit, such as a chapter, scene, or section.
  2. Draft a linked beat sheet that describes the intended turns, promises, reveals, examples, or page beats.
  3. Ask Scritorio to critique the beat sheet for structure, tension, clarity, argument flow, or reader expectation.
  4. Draft prose in the linked manuscript file.
  5. Compare prose against the beat sheet to identify missing turns or drift.
Beat sheets should remain planning documents by default. They may inform coaching and editorial context, but they should not be assembled into the book unless the author explicitly chooses that.

Coaching Assignment Loop

The writing coach should support a classroom-like practice loop for one-off coaching and guided curriculum modules:
  1. Choose a skill area, manuscript passage, or curriculum path.
  2. Generate a focused assignment with a teaching note, prompt, constraints, and rubric.
  3. Show the assignment in the project navigation.
  4. Let the author draft a response inside the assignment.
  5. Submit the response to the coach for review.
  6. Save rubric feedback and one focused revision prompt.
  7. Let the author revise, complete, archive, or connect the assignment to a manuscript task.
Assignments should be durable Markdown project objects. They may begin from a chat exchange, but they should not disappear into chat history. In curriculum mode, this loop repeats across an ordered module sequence. The app should show the active module, current assignment, completed assignments, and next recommended step.

Fiction Editorial Loop

For fiction, Scritorio should support:
  • copy edit
  • line edit
  • developmental review
  • canon review
  • continuity review
  • character review
  • blind persona reader reaction
  • managing editor summary
  • revision coaching

Nonfiction Editorial Loop

For nonfiction, Scritorio should support:
  • argument review
  • structure review
  • claim and source audit
  • clarity review
  • audience fit review
  • teaching-value review
  • visual opportunity review
  • managing editor summary
  • revision coaching

Visual Development Loop

For diagrams, infographics, timelines, maps, and explainers:
  1. Select manuscript text or notes.
  2. Ask Scritorio to identify visual opportunities.
  3. Choose a visual type and purpose.
  4. Preview source context and prompt.
  5. Generate or revise the visual through a provider.
  6. Save the prompt, result, source notes, and revision history locally.
  7. Link the asset back to the manuscript section it supports.

Report And Assignment Storage

All AI review output should be saved as Markdown reports. Coaching assignments, submissions, feedback, and revision notes should be saved as Markdown assignment files. Visual assets should be saved as local files with Markdown sidecars for prompt, source context, author notes, and revision decisions.

Agent-Operated Workflows

Agents should be able to operate Scritorio through the CLI without needing to drive the UI. The ideal agent loop:
  1. Inspect project state.
  2. Preview the context for a requested action.
  3. Ask the author for approval if content will be sent to an external provider.
  4. Run the action through the CLI.
  5. Save reports or assets locally.
  6. Return structured results and file paths.
This makes Scritorio useful inside agent workflows while preserving author control.