code0Vörr AIContactDocs
Book a Demo

How to Give ChatGPT Context About Your AWS Infrastructure

Ask ChatGPT "why is my service slow?" and you get a generic checklist. It doesn't know which services you run, how they're connected, or what changed yesterday. This article covers the ways to give ChatGPT real context about your AWS infrastructure, from copy-paste to a live connector, and how to do it without handing over credentials.

Option 1: Paste it in

Export a CloudFormation or Terraform file, a security group or some log lines, and paste them into the chat.

Good for
a one-off question about one resource.
Limits
it's manual and quickly out of date. You only share what you already suspect is relevant, and there's a real risk of pasting secrets.

Option 2: Describe your architecture

Write a summary of your services and how they connect, and reuse it as a prompt or in project instructions.

Good for
general design discussions.
Limits
it's from memory, so it misses what you've forgotten or don't know, which is often where the problem is.

Option 3: Connect a live context source

ChatGPT plans that support custom connectors can retrieve facts from external systems during a conversation, using MCP. Instead of you describing AWS, ChatGPT looks up the real state.

What to look for in a connector:

  • Read-only. It shouldn't be able to change anything.
  • Credentials stay out of ChatGPT. The assistant gets facts, not keys.
  • More than AWS. Infrastructure questions usually involve code, deployments and monitoring too.
  • Sources you can check. Every fact should say where it came from.

Setting up Vörr as ChatGPT's AWS context

Vörr connects your AWS accounts, repositories, observability tools and CI into one engineering context, and exposes it to ChatGPT through a single read-only MCP connector.

  1. Connect AWS in Vörr. Use an access key or an assumed role with read-only credentials, give the account a friendly name and choose which workspaces can use it.
  2. Start a scan. Connecting doesn't scan automatically. Start the scan when you're ready, and Vörr maps your supported resources.
  3. Connect your repositories (GitHub, GitLab or Bitbucket) so answers can point to the code behind the infrastructure.
  4. Get the connector URL. In Vörr, open Chat and choose Connect with ChatGPT to see the steps and the MCP URL.
  5. Add it in ChatGPT using the custom connector setting, then approve access on Vörr's consent screen.
  6. Check it works. Ask about a service you know exists. The answer should show your real resources, with their sources.

Questions to try

  • "What is the checkout service exposed to, and what does it depend on?"
  • "If this database goes down, what breaks?"
  • "Why did the latest payments deployment fail?"

FAQ

Does ChatGPT see my AWS credentials?

No. Authorizing the connector doesn't expose Vörr's stored cloud, Git or observability credentials to ChatGPT.

How do I remove access?

Remove the connector in ChatGPT's settings. That revokes access and changes nothing inside Vörr.

Does it work with Claude too?

Yes. Vörr connects to both Claude and ChatGPT.