S3 and SFTP storages

A storage connection gives QueryLane access to files in S3-compatible object storage, such as AWS S3, Yandex Object Storage, Cloud.ru or MinIO, or on an SFTP server. You browse folders, download and upload files, open tabular files as tables and query them with SQL. Jobs read and write the same files with the Write file and Transfer files steps.

Add a storage connection

1. Click Connections in the left rail.
2. In Data sources, click + (Add source).
3. Under File storages, choose S3-compatible storage or SFTP.
4. Enter a Connection name and fill in the fields described below.
5. Click Test, then Create.

Storage connections appear under Storages in Data sources. Keys and passwords are kept in the encrypted vault of the profile, like database passwords, see Where passwords are stored.

S3-compatible storage

• Endpoint: the address of the service, for example https://storage.yandexcloud.net
• Region: the region used to sign requests, us-east-1 by default
• Bucket: optional. Without it, the connection lists every bucket the key can see
• Folder: optional folder inside the bucket where browsing starts, for example exports/
• Access key ID and Secret access key
• Path-style addressing: puts the bucket in the path instead of the host name. Most self-hosted stores and MinIO need it, so it is on by default

Session token is for temporary keys only. Skip TLS certificate check is meant for a server with a self-signed certificate you trust.

With a Bucket, the connection reaches only that bucket, and with a Folder only that folder inside it. Explorer, queries, job steps and agents all stay within these limits. A Folder needs a Bucket.

SFTP

• Host, Port (22 by default) and User
• Sign in with: Password, Private key with its Key passphrase, or SSH agent
• Start folder: where browsing starts on the server, / by default. The connection reaches only this folder and what is inside it
• Host key fingerprint: the SHA256 fingerprint the server must present. When it is empty, QueryLane checks the server against the known_hosts file

Browse files

Click a storage connection to expand it. S3 shows buckets, then folders and files; SFTP starts with the folders of Start folder. Large folders load page by page, with Show more at the end of the list.

Double-click a folder to open it in a tab with Name, Size and Modified columns and a path you can click to go up. Right-click a file or folder for its actions:

• Open and Open folder
• Download: saves the files to a folder you choose. QueryLane asks before it replaces files with the same names
• Upload: sends one or more files into the folder, and asks before it replaces files that are already there
• New folder and Refresh
• Query: opens a query tab over a tabular file, see Query files with SQL
• Copy address: copies the address of the file, such as s3://reports/2026/10/orders.csv
• Delete: asks first. A folder is deleted with everything inside it

Downloads and uploads show their progress and can be stopped. A stopped transfer leaves no half-written file behind, and an S3 download that loses its connection continues from where it stopped.

On S3 a file also has Share link, which creates a download link that works for 1 hour, 1 day or 7 days. Anyone with the link can download the file until it expires.

Open files as tables

CSV, TSV, JSON, JSONL, Parquet, Excel (XLSX and XLS) and XML files open in a table tab, like a database table. Paging, sorting, filters, the Table, JSON and Tree views and Export work the same way, see Results and editing. Gzip-compressed CSV, TSV, JSON, JSONL and Parquet files open too. The data is read-only.

A bar above the table shows the size and date of the file, with buttons to download it, copy its address or share link and delete it. When an S3 bucket keeps versions, Versions in the bar lists them, and each version can be downloaded or removed with Delete this version. For a workbook, choose the Sheet and the Header row, the row that holds the column names.

A large file is prepared once: QueryLane says The file is not prepared for viewing yet, and Prepare makes a local copy, with progress in the bottom-right corner. After that, pages, counts, sorting and filters are quick. Prepared copies are kept for 7 days after their last use, up to 4 GB in total.

Text files open as text with Start of the file. Other files show No preview for this file type and can be downloaded.

Query files with SQL

A query tab on a storage connection runs DuckDB SQL, where a file is named as a table by its address in double quotes:

SELECT status, count(*) AS orders
FROM "s3://reports/2026/10/orders.parquet"
GROUP BY status
ORDER BY orders DESC;

On SFTP the address starts with sftp:// and the absolute path on the server, such as "sftp:///exchange/partners.csv". For a workbook, add the sheet and header row after #: "s3://reports/budget.xlsx#sheet=Plan&header=3".

Query in the menu of a file opens a tab with a first query over it, and Copy address gives you the address to paste. SQL steps in jobs can use a storage connection as their Data source in the same way.

Agents and storages

Storage connections have the same Agents block as databases, see Agent access. With Read only, agents can list folders, read files and run SQL over them. With Read and write, they can also write and delete files, and Approve changes decides whether that waits for you. Uploading a file from your computer always asks first.

Agents work only within the bucket, folder or start folder of the connection, and share links are created only in the app. See Connect your agent.