> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cortex-labs.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Set up ingestion with Kato

> Configure a supported source and ingestion job, then choose a bounded run and inspect its results.

Ask Kato to configure collection into an existing dataset, then start the agreed work and inspect the ingestion record. Use this guide when you know which material to collect and have authority for the source and destination.

## Gather the inputs

You need the relevant Vantage, workspace, dataset and job permissions, permitted source access, and an authorised execution principal. Prepare the source URL or identifier, the target dataset and collection, required connection settings, exact time or input scope, and usage limits.

Use your own source and dataset identifiers. Check that the connector supports the requested input and limits. If time matters, choose exact dates and a time zone from your task and confirm how the source applies them. A web page fetched today may contain older content without supporting a date-filtered collection request.

Check the bundle's declared secret names and status. An Admin sets tenant secrets in **Console → Secrets**. Give Kato the name, never the password, API key or token value. Secrets have no workspace scope.

## Request configuration first

Send a request such as:

> Configure ingestion for \[collection name] in \[workspace name] using \[my source URL or identifier] and \[target dataset]. Check the supported connector and source scope. Propose the source connection and job, including the time filter or URL/input limits, principal and expected usage. Do not start collection, schedule it or enable automatic enrichment yet.

Replace the bracketed fields with your actual identifiers and limits. Ask Kato to name any gap between your requested scope and what the source supports. If the source is unsupported or credentials are missing, resolve that dependency before asking for a run.

After checking the proposal, ask Kato to create or revise the specified configuration. State whether it should reuse an existing source/job. Keep unrelated dataset settings and existing triggers unchanged unless you intend to change them.

## Inspect the saved job

Open the resulting dataset and its ingestion-job controls. Check the job's source, target, parameters, credentials reference and enabled state. Confirm that the target dataset is attached through the intended collection.

Inspect any existing automatic processing on the dataset. Collection can feed processing already enabled there. Asking for no new enrichment configuration does not pause an existing binding.

If Kato completed only part of the configuration, follow [Recover an incomplete Kato task](/kato/recover-task). Keep the identifiers of completed resources before retrying.

## Start the agreed work

For a controlled first run, request one manual collection with the supported limits you agreed:

> Run the identified ingestion job once with the agreed source scope and limits. Do not add a recurring trigger. Return the run identifier and status so I can inspect it.

Start any enrichment separately, specifying the selected published version, dataset binding and manual or automatic processing choice. [Develop and test an enrichment with Kato](/kato/develop-enrichment) explains that route.

For custom ingestion code, inspect an isolated test run and its output/checkpoint proposals before activation. A custom production start or trigger requires a successful test of the same job/version. System bundles have their own publication review and need no per-job test. Test output and checkpoints stay separate from production. Kato assistance does not remove source, credential or execution requirements.

## Check collection and processing

Follow the returned work record with [Find resources and track work with Kato](/kato/find-and-track-work). Check accepted input, failed or skipped items, extraction and downstream processing separately.

| What you see | What to check |
| - | - |
| Empty collection | Source access, requested scope and available input. Empty results do not establish that no relevant events occurred. |
| Accepted files, unfinished processing | Which content is searchable or ready for the intended enrichment. Collection completion alone is insufficient. |
| Partial or failed work | The affected input and error, what was retained, and what a retry would repeat. |
| Completed work | Actual coverage, source versions and usage against the agreed limits. |

Finish by opening the destination dataset and checking the available evidence. Use [Organise data through collections](/vantage/set-up-a-workspace/organise-collections) for workspace reach and [What Kato can do](/kato/capabilities) for operation requirements.

For detailed configuration, use [Choose a data source, dataset, and collection](/vantage/bring-in-data/choose-source-dataset-collection), [Run ingestion and inspect processing](/vantage/bring-in-data/run-ingestion) or [Build and test a custom ingestion job](/vantage/bring-in-data/custom-ingestion-job).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.