Use InfraNest from an AI assistant

Connect an AI assistant like Claude, Cursor or VS Code to InfraNest so you can ask questions about your infrastructure in plain language, instead of clicking through pages.

Overview

• Ask things like "Which monitors were down last night?", "Where does mail.example.com point?" or "Which domains expire this quarter?" and get answers pulled straight from InfraNest.
• The assistant connects using an API token you create yourself, and it only reads data — it can never change, add or delete anything.
• It also never sees secrets such as domain transfer codes, certificate files or tokens, even if the token it uses could technically read them.

Create a token for the assistant

1. Go to My account, then open Developer.
2. Select New token.
3. Give the token read access only to the areas you want the assistant to see.
4. If you'd like to try it on sample data first, select Create a sandbox and use that instead of your live organisation.
5. Select Create to finish.

Connect the assistant

1. From My account → Developer, open Set up an AI assistant.
2. On that page, find Use InfraNest from an AI assistant.
3. Follow the link to the developer site for exact setup steps for Claude Code, Cursor, VS Code or Claude Desktop: https://developers.infranest.io/ai-assistants.

4. Follow the instructions there to paste in your token and finish connecting.

NoteIf you use claude.ai or ChatGPT, you don't need a token — see the separate guide for connecting those directly.

Stop an assistant

1. Go to My account, then open Developer.
2. Find the token the assistant is using.
3. Select Delete (or revoke the token).
4. The assistant immediately loses the ability to read anything from InfraNest.

Tips

TipStart with a sandbox token so you can see how the assistant behaves before pointing it at your real infrastructure.

Troubleshooting

• The assistant can't see something I expect — check that the token has read access to that area, and that your organisation's IP allowlist permits the address the assistant is connecting from.
• I'm not sure what the assistant is doing — every request it makes counts towards the token's rate limit and shows up in the audit log under the token's name.
• The assistant gave me a wrong-sounding answer — remember text in your own data (a DNS record, a monitor name, an incident note) can be written by anyone on your team. Double-check before acting on it.