Nodes

Document node

Put PDFs, Office files and text files on the canvas so connected chats can read them.

A Document node turns a paper, report, slide deck or spreadsheet into a source the AI can read. It sits next to the conversations about it, so you can see which chats drew on which source.

Add a long report once, connect it to three chats, and each can work through it from a different angle: one summarizes, one checks the numbers, one drafts a reply.

Add a document

  • Toolbar: press U or D (or click Upload), then pick one or more files. Each one lands near the middle of your view: images as Image nodes, everything else as Document nodes.
  • Right-click menu: right-click the canvas and choose Document Node (marked Beta).
  • Drag and drop: drop files from your file manager onto the canvas.
  • Paste: copy a file, then paste it onto the canvas.

Files Slashspace can't use show a message such as "File type .zip is not supported".

Supported files

KindExtensions
PDF.pdf
Word.docx, .doc
PowerPoint.pptx, .ppt
Excel and CSV.xlsx, .xls, .xlsm, .ods, .csv, .tsv
Books.epub
Text and Markdown.txt, .md, .rtf, .html, .json, .xml
Code and config files.py, .js, .ts, .sql, .yaml and many more

Keep documents under 50 MB. The file picker skips anything larger.

How reading works

When you add a document, Slashspace reads it on your computer and converts it into Markdown (plain text with headings and tables). Nothing is uploaded to do this, so it works offline and without signing in. Scanned pages go through text recognition on your computer, which reads English. Text and code files need no converting: chats read them straight from the file.

The results are saved in a files folder inside the canvas's folder, with a simple PDF preview for Office and text files.

Two things you choose can send it further. A chat connected to the document sends its text to that chat's model (or the PDF itself, if reading hasn't finished), and a canvas in the cloud syncs the original file. See Privacy and your data.

The node shows its progress along the bottom:

StatusMeaning
Reading document... or Reading book...Slashspace is reading the file. Larger files and scanned pages take longer.
The page preview, with no statusDone.
A red message with RetryReading stopped. See If a document can't be read.

PDFs show their first page right away, with a page count. Spreadsheets show as a table. Other files show a file-type card until reading finishes. Double-click the preview to read the whole document in the side view.

Documents from older versions

Versions up to 4.1.0 uploaded documents and converted them on Slashspace's servers. Later versions read those documents again on your computer when the original file is still there. A document that would lose scanned text this way keeps its cloud copy. Uploaded copies aren't deleted automatically: see Deleting your data.

Open Markdown and Open PDF

Once a document is read, the side view shows Ready, with buttons that open its converted copies in your computer's default app:

  • .md (Open Markdown): the converted text. This is what connected chats read, so check here that a table or slide came through.
  • .pdf (Open PDF): a plain PDF of that text, for Office and text files. For Word and PowerPoint files, the node preview and side view show this version, so the original layout and images don't carry over.

On an older document that still has a cloud copy, these buttons open a public link in your web browser instead.

Use a document as context

Connect the Document node's right handle to a chat node, or type @ in a chat and pick the document. Connections and context explains how context flows. A chat in Isolated context ignores its connections, documents included.

Once a document is read, a connected chat receives the full converted text with every message. Nothing is summarized or trimmed to snippets.

Wait for reading to finish before you ask. Until then, Office files are left out, and the chat tells you the file "is still being processed, so it wasn't included in context". A file that can't be read gets a similar notice, and the chat still answers.

Very large documents

  • If the converted text is larger than about 500 KB (very roughly 80,000 words), the chat leaves it out and tells the model it was omitted. Split long books or reports into smaller files, such as one per chapter.
  • The whole document is sent again with each message, so connect big documents only to the chats that need them. This leaves more of the model's context window for the conversation.
  • A model reading a document through its canvas tool, as Claude Code and Codex can, gets up to 64 KB of its text. Orchestrator mode sees the document's name and location, not its text.

If a document can't be read

The node shows a red message and a Retry button.

  • "The file was moved or deleted." Put the file back and click Retry, or add it again.
  • "Slashspace can't read files in this location. Drop the file on the canvas again." Dropping or picking a file gives Slashspace access to it.
  • "This file is password-protected, so it can't be read." Remove the password and add the file again.
  • "This file looks damaged and couldn't be read." or "This file is too large or complex to read." Export it as a PDF, or split it, and add that instead.
  • "This file has no saved copy on disk, so it can't be read." It was pasted with no file behind it. Save it to your computer and add it from there.
  • The node never finishes reading. Delete the node and add the file again.
  • A Word or PowerPoint file keeps failing. Export it as a PDF and add that instead.

Still stuck? See Troubleshooting and support.

On this page