BrunelDocs

Guides

Infrastructure from your AI agent

Use Brunel from Claude Code, Cursor or VS Code through MCP. Analyze the repository you are working on, design a priced architecture and write validated Terraform into your project from a chat.

Your coding agent knows your code; Brunel knows cloud architecture, prices and validated templates. Connected through MCP, the agent can design your production infrastructure without leaving the editor.

Set up (two minutes)

  1. Create a read and write key in Platform → AI agents; it fills the configuration for you.
  2. Add the server to your client. For Claude Code:
bash
claude mcp add --transport http brunel https://api.brunel.cloud/mcp \
  --header "Authorization: Bearer brn_w_…"

Other clients: Connect your client.

Prompts that work

  • "Analyze this repository with Brunel and summarize what it runs."
  • "Design its production infrastructure with Brunel for a €500 monthly budget on AWS, with EU data residency."
  • "Why does the Growth tier include a NAT gateway? Remove it if we can."
  • "Write the Growth tier's Terraform from Brunel into infra/."

What happens behind the scenes

The agent calls analyze_repository, then design_infrastructure with the context you gave, refine_architecture for changes and get_files for the Terraform and CI. Prices and files come from Brunel's catalogs and templates, never from the agent's own guesses, and every call appears in your request logs.

Keep it safe

  • Give the agent its own key with an expiry; revoke it independently.
  • Private repositories work with the GitHub access of the session that created the key.
  • Review the Terraform like any other code change before applying it.
Infrastructure from your AI agent · Brunel Docs