Preview a file before saving it to Memory

Not every file you touch in a flow deserves a permanent home in Memory. Sometimes you just want to glance at what’s sitting in a supplier’s SFTP drop folder, or sanity-check today’s export before deciding whether it’s worth keeping. This guide shows how to fetch a file and preview it inline — text, image, or PDF — without creating a document.

The building blocks

Three steps work together, and each does exactly one job:

• List files (storage) — browses a connected storage server (SFTP, WebDAV, FTP, S3, SMB) and returns names, paths, and sizes. No file content.
• Read file (storage) — pulls one file’s bytes onto local disk and returns file_path (plus filename, size, mime_type). This is a pure fetch: it never writes to Memory or the document store.
• Send pop-up — shows a modal to whoever is watching the run. Its attachment field can render another step’s fetched file inline.

Both storage steps are currently in Beta.

Fetching is not saving

Read file (storage) intentionally stops at file_path — a temporary local copy. If you want the file to persist in Memory (reviewable, searchable, entity-linked), chain a Document Store step after it. Skip that step entirely and the file is only ever used transiently by the steps that read it.

Wire it up

1. Add List files (storage) (or skip straight to step 2 if you already know the path).
2. Add Read file (storage), giving it the file’s path. Label it something short — the label is how you’ll reference it next, e.g. read_step.
3. Add Send pop-up. Set:
   • Title — a short headline, e.g. Content of the file.
   • Attachment — the label of the step that fetched the file, e.g. read_step. This is not a template — don’t write {{ read_step.file_path }} here, just the plain label read_step. The pop-up looks up that step’s output and serves the file itself.

Run the flow. The pop-up shows your title, any body text, and — inline — the fetched file’s content.

What renders, and what doesn’t

The attachment renders based on the file’s type:

File type	What you see
Text (.txt, .csv, .json, …)	Content shown inline
Image (.jpg, .png, …)	Image shown inline
PDF	PDF shown inline
Anything else	A download link — never an error

If you decide to keep it

Add a Document Store step after Read file (storage), wired to the same file_path. That’s the only difference between “just looking” and “keeping it in Memory” — the fetch itself never changes.

Where to go next

• Connected Storage — the five protocols, how naming multiple connections works, and connecting without a password.
• Send pop-up — the full field reference, including the attachment field’s exact behaviour.
• Running & debugging a flow — the step log shows resolved inputs, useful when a file isn’t rendering the way you expect.
• Keep a human in the loop — for when the person looking at the file needs to make a decision, not just glance at it.