> ## 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.

# Meet the agents

> Vapr does not write code or run pipelines itself. It routes you to the right specialist, then takes the conversation back.

* **Vapr**: Keeps the conversation and the state of your application. It suggests each next step and waits for your answer.
* **Coding Agent**: Builds the application. It gathers your requirements, asks about technical details, writes the code and runs the test cases, all inside the application's own dev container.
* **Pipeline Agent**: Deploys the application. It asks for the values the deployment needs, such as connection details and topic names.
* **Testing Agent**: Proves it works end to end. It simulates messages through the pipeline, inspects what comes out and checks it against what you expected.
* **Monitoring Agent**: Watches the running application. It pulls logs and observability data and answers in the same chat.

> [!NOTE]
>
> Each agent works only on its own phase, with only the tools and context that phase needs. When it finishes, it hands control back to Vapr, and you can ask any agent a follow-up in the same thread.

For questions about Kafka, Kubernetes and monitoring that are not tied to building an application, Vapr uses the existing Kafka, Kubernetes and Grafana Monitoring agents. See **AI Agents**.


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