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

# Deploy input / output connectors

## Input / Source Connector

Choose a Universal or Industry input connector, or build a custom one to source the data. Configure, deploy and maintain them with ease

<video src="https://mintcdn.com/zeliot/YbmUp2z2-EHEiH1u/videos/Source.mp4?fit=max&auto=format&n=YbmUp2z2-EHEiH1u&q=85&s=886a537350e72ef9a7e88663db8d0023" controls />

## Destination / **Sink Connector**

###### **Choose a universal output connector or build a custom one to sink the data. Configure, deploy and maintain them with ease**

<video src="https://mintcdn.com/zeliot/YbmUp2z2-EHEiH1u/videos/Sink.mp4?fit=max&auto=format&n=YbmUp2z2-EHEiH1u&q=85&s=83eb33ed7f1c6580ab246561ed21e9dc" controls />

## Choose a saved connection and topic (New in v3.0.0)

For every input and output stream setting of a service, the deployment panel asks for two choices.

1. **Choose the broker.** The list holds the connections in your registry, such as Condense Embedded NATS, Condense Kafka or Confluent Cloud.
2. **Choose the topic or subject.** The list is fetched from that broker with your credentials, so it shows only what you are allowed to use.

### The access check

Next to each choice a badge tells you, before you deploy, whether your credentials have the access the stream needs: read for an input, write for an output.

* **Fetching**: Condense is checking the permissions and loading the topics.
* **Verified**: the badge says that read access (for an input) or write access (for an output) is verified.
* **Failed**: your credentials lack the permission. Ask the owner of the broker to grant it.

Select **Refresh** to reload the topics and permissions after something changes on the broker.

### Mix brokers

An application can read from one broker and write to another, for example consume from Embedded NATS and publish to Confluent Cloud. Choose a connection for each stream.

### Why it matters

* Nobody types a raw secret during deployment. The connection supplies it.
* Typos in topic names cannot happen, because you pick from a list.
* A missing permission shows before you deploy, not as a failure afterwards.

You can also set a title for the instance, up to 40 characters, and choose the worker size in the same panel.


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