DocsWriting Studio › Add context: @file, @folder, selection, paste

Add context: @file, @folder, selection, paste

Last updated June 4, 2026 · 3 min read

By default the Coach has read the whole book and treats it evenly. Want to spotlight Chapter 3 plus character notes? Four ways.

The Slima AI Coach panel: the message box accepts @file / @folder context, pasted text, or your current selection

Four ways

1 · Add a file

Drag a file from the left file tree straight into the message box — or right-click the file and choose Add to Chat. It becomes a context chip above the input.

The Coach treats chipped files as priority context — weighted above the rest of the book.

You can add several chips, then type your question:

[Chapter 3] [Protagonist] — check whether J. Doe's reaction makes sense

To drop a chip, click it and press Backspace, or use its remove (×) button.

2 · Add a folder

Drag a folder into the message box the same way. Every file inside the folder joins priority context as a single chip.

Good for "I want all character notes weighted up this round."

3 · Selection

Highlight text in the editor → the Coach panel shows "N characters selected".

Ask a question, and the AI treats that selection as highest-priority context:

Selected: "Lighthouse keeper J. Doe had worked the reef station for thirty years."
You ask: "smooth this out"
The AI smooths that sentence.


4 · Paste

Paste text directly into the message box (or drag a file in):

  • Text → treated as "this is what we're discussing / revising"
  • File (.md / .txt / .docx) → parsed and added as context

Good for external content (a chapter a friend sent, someone else's draft).


Priority layers

Four priority levels in the Coach's context:

1. Selection / paste  (highest — the active object)
2. File / folder chips (priority reference)
3. Writing Guidelines.md    (behaviour rules)
4. Whole book         (background)

Higher overrides lower on conflict.


Limits and best practice

  • Per-message context capacity is bounded (depends on the current AI model) — adding every file as a chip at once may get compressed
  • Long chats auto-compress old context — pin or summarise important conclusions (Advanced behaviours)
  • Edits to a file refresh context immediately — @ Chapter 3, edit Chapter 3, the next message sees the new version

Related

Was this helpful?