Skippr Docs

Appearance

How to Define SFTP Ingestion Scope Clearly ​

July 2026

An SFTP source is only clear when the team can name the remote files the connector should pick up, not just the server it can log into.

Short Answer ​

Define SFTP ingestion scope with the host, the authenticated username, and, most importantly, the exact remote_path or glob. The source docs make that field required because the source boundary is the file set the connector can list and read, not the SSH account in the abstract.

That means remote_path: "/exports/orders/*.json" is a clear contract while remote_path: "/" is almost never one. A dedicated automation user can help, but the real shape comes from the directory or pattern. Since the namespace is sftp.{filename}, file naming also becomes part of how downstream systems recognize what landed.

Why Teams Struggle with This ​

SFTP integrations become harder than they should be when teams prove login success and stop there. In production, the important question is which files that login will pull tomorrow morning and whether those files still represent one coherent feed.

  • An SSH account with broad directory access can make the first ingest boundary much larger than intended.
  • A loose remote_path can quietly pull test files, archived files, or unrelated partner drops into the same feed.
  • File naming conventions matter because downstream consumers often infer the landed dataset from the filenames they see.
  • Choosing password or key auth affects security posture, but the remote path is what defines the dataset boundary.

How Skippr Handles It ​

Skippr keeps SFTP scope visible because the connector surface stays literal: host, port, username, auth method, and remote path. That makes review more concrete than a generic statement that "the SFTP connector is set up."

The best starting point is usually one directory or one glob for one business feed. If a vendor publishes orders_2026-07-19.json and returns_2026-07-19.json in different directories, the connector boundary should reflect that separation rather than collapsing them into one broad pickup.

  • Uses remote_path or a glob as the actual file boundary.
  • Supports SSH key auth or password auth without changing how scope is defined.
  • Keeps host and path selection readable in config for production review.
  • Produces file-based namespaces that stay tied to the selected pickup pattern.

What the First Useful Version Looks Like ​

The first useful version is one SFTP host, one automation identity, and one remote path pattern the team has validated against real files. That gives you a stable file contract with minimal ambiguity.

If the directory still contains several unrelated file families, split the pickup early or negotiate a cleaner upstream drop location. SFTP pipelines are simplest when path layout already matches the business boundary.