CSV Intake Agent
The CSV Intake Agent connects CSV files to DCH. You upload a file, and the agent stores it, reads its header row as the column list, and makes its rows available to DCH as a data source.
Like every agent, it is built on the Intake Agent Core and shares the Home, Data Preview, and Data Jobs screens described there. It adds one screen of its own, CSV Files, where all of the work happens.
Each uploaded file becomes one configuration with one source entity. The entity is named after the file, without the .csv extension, and its properties are the file's columns. There is nothing to wire up: uploading the file creates the configuration, and deleting the file deletes it again.
The CSV Intake Agent is open-sourced on GitHub under the GPL-3.0 license.
CSV Files
The CSV Files screen lists every uploaded file with its File Name, the Configuration Name DCH knows it by, its Size, and who created and last changed it (Created By, Updated By). The refresh button reloads the list.

Each row has three actions: Preview, Edit, and Delete.
Previewing a file
The Preview action (the eye icon) opens the file as the agent parsed it. You see the file name, its size, its total row count, and when it was uploaded, followed by the full list of CSV Headers and up to ten rows of data.

This is the quickest way to confirm that the agent read your file the way you expected, in particular that it picked the right delimiter and found the header row you intended.
The preview is parsed fresh from the stored file each time you open it, so it is never stale. This is separate from the shared Data Preview screen, which previews configurations across the agent.
Editing a configuration
The Edit action opens the configuration behind the file.

- File Name: Read-only. To replace the file, delete this one and upload the new one, or upload it under a different configuration name and slug.
- Configuration Name: How the configuration appears in DCH. Must be unique.
- Configuration Slug: The stable identifier DCH refers to it by. Must be unique, and may contain only lowercase letters, digits, and hyphens. It cannot start or end with a hyphen.
- Batch Size: How many rows the agent sends to DCH per batch during an import. Defaults to
10000. - Enabled: Whether the configuration is offered to DCH. Disabled configurations are also hidden from the shared Data Preview screen.
Name and slug are only checked when you press Save, and both are matched exactly. Two configurations may therefore differ by capitalization alone (Orders and orders), though their generated slugs would collide.
Deleting a file
The Delete action asks you to confirm, naming the file, and then removes both the stored file and the configuration built from it. DCH can no longer import from it.
Uploading files
Click Upload on the CSV Files screen. Choose up to ten .csv files at a time, up to 100 MB each.

Every selected file gets a Configuration Name and a Configuration Slug, generated from the file name: quarterly_orders.csv becomes the name Config_quarterly_orders and the slug quarterly-orders. Both are editable before you upload, and the red cross removes a file from the selection.
Upload Files then processes each file in turn and reports any that failed. A file is rejected if:
- Its configuration name or slug is already taken. The name is checked first.
- Its slug is not a valid slug.
- It is empty, or has a header row but no data rows.
If some files fail, the upload window stays open and lists the errors, while the other files are already uploaded. Remove those from the selection before you try again, otherwise they are rejected as already existing. The list behind the window does not update by itself, so use the refresh button after closing it.
Uploading a file whose name matches an existing configuration does not replace that configuration, it is rejected. Delete the old file first, or give the new one a different configuration name.
How files are parsed
The agent makes a few decisions for you when it reads a file, and it is worth knowing which.
Delimiter: Detected automatically, by comparing how many commas and semicolons appear in the first few lines. Semicolon wins only if it is strictly more common; otherwise the agent uses a comma. No other delimiter is recognized, so tab-separated and pipe-separated files are not supported.
Header row: The first row is always the header. The agent cannot tell whether it really is one, so in a file without a header, the first row of data becomes the column names and is not imported.
Encoding: UTF-8, with a byte order mark detected if present. There is no setting to override this.
Quoting: Standard CSV quoting. Quoted fields may contain the delimiter, and may span multiple lines.
Types: None are inferred. Every column is imported as text, whatever it looks like.
The upload does not check the columns or the contents of each row, so check these yourself:
- Columns with the same name, including several empty ones, are accepted. An import keeps only the first column of that name and the others are lost, while the file's preview shows the last one.
- Rows with too few columns, or with incorrect quoting, are accepted. The upload succeeds, but reading the file fails when the import or the file's preview reaches such a row.
- Rows with too many columns are accepted. The extra values are ignored.
Data Preview
The CSV Intake Agent supports the shared Data Preview screen, which lists the configurations that are Enabled. Use the file-level Preview action described above when you want to inspect the file itself.