> For the complete documentation index, see llms.txt.
Skip to main content

Check out Port for yourself ➜ 

Getting started

This walkthrough shows how to build an agentic workflow that provisions a new cloud resource using Infrastructure as Code (IaC). A developer or agent expresses the intent to create a resource, and the workflow fetches context, generates an implementation plan, creates a resource request, and opens a pull request against your GitOps repository.

See it live

Explore a live implementation of this workflow in the public demo.

Prerequisites​

  • A GitHub integration connected to the GitOps repository that holds your IaC.
  • A coding agent with access to the GitOps repository, for example the Cursor Agent node.
  • A Slack integration for notifications (optional), for example to send a Slack message when the pull request opens.

How the workflow works​

Request cloud resource workflow: self-service trigger, fetch relevant context, generate implementation plan, create resource request, and open a pull request

The workflow chains the stages of a golden path into a single agentic flow. Each stage passes its output to the next.

  1. Trigger with a self-service form. The workflow starts from a self-service trigger that collects the resource details, for example resource type, service, and environment. An agent can fill this form from the IDE through the MCP server, or a developer can fill it from the UI.
  2. Fetch relevant context. The workflow pulls context from the software catalog: existing resources, the target service, its owner, and the Terraform modules and GitOps templates that define your standards.
  3. Generate an implementation plan. An AI action node uses that context to generate an implementation plan, describing exactly which IaC changes are needed to provision the resource within your golden path.
  4. Create a resource request. The workflow records the request in your catalog, so the resource and its lifecycle are tracked from the moment it is requested.
  5. Open a pull request. A coding agent, such as the Cursor Agent node, applies the planned IaC changes and opens a pull request against your GitOps repository using the GitHub integration action. Your existing GitOps pipeline then provisions the resource once the PR is merged.

Possible enhancements​

Once the base workflow is running, you can extend it:

  • Add skills to the AI agents. Attach skills to the agents in the workflow so they follow your organization's specific "how-to" for each resource type.
  • Conditional auto-merge. When the requested change carries no risk, based on cost, environment, or policy, auto-merge the pull request instead of waiting for manual review. Higher-risk changes still route to a human-in-the-loop approval step.

Next steps​

  • Golden paths - understand golden paths as the "how-to" of your organization, across the full resource lifecycle.
  • Build workflows - learn how to build and test workflows in Port.