From a chat to a live app#
You built an app with Claude or Claude Code, and it runs on your computer. To put it online, give Nimbus Land the folder or the repository:
- From your computer: install the
nimbuscommand-line tool and runnimbus deployin the project's folder. It uploads the folder, builds it, waits until it is live and prints its address. - From GitHub: push the project to a repository, choose New on Applications in the console, pick the repository and deploy.
Nimbus Land works out the language and framework, shows a deploy plan before anything builds, and fills in what a first deploy usually needs. Your first app walks through both paths.
Connect Claude to your apps#
The Nimbus Land connector lets Claude see and operate your apps from a conversation, on claude.ai, in Claude Desktop and in Claude Code. It is a remote MCP server at https://nimbusland.ca/mcp. You sign in with your Nimbus Land account on a consent page; no API key is copied anywhere.
If Nimbus Land appears in Claude's connector directory, choose Connect there. Otherwise add it by URL under Settings → Connectors → Add custom connector.
Once connected, Claude can, with your approval and within your role:
- list your apps and read their status, builds and build logs;
- read runtime logs, errors, traces, traffic and analytics;
- set environment variables, without ever seeing secret values;
- deploy from the connected repository, rebuild and run one-off commands;
- run read-only queries against your managed Postgres databases.
Claude deploys from the app's connected repository or a Git URL; to deploy files that exist only on your computer, use nimbus deploy. Using Nimbus Land from Claude lists every tool and its limits.
Claude Code#
In Claude Code, add the connector with one command:
claude mcp add --transport http nimbus https://nimbusland.ca/mcpThen run /mcp and choose nimbus to sign in. The Nimbus Land Claude plugin adds the same connector plus a skill that teaches Claude Code the nimbus command-line tool, for work that needs your local project.
What goes to Anthropic#
Claude runs on Anthropic's infrastructure, outside Canada. Everything the connector returns (app names, statuses, log lines, build logs, query results) becomes part of your conversation with Claude. Secret values are never returned. Your app and its data stay hosted in Canada; only what you ask Claude to read reaches Anthropic. See Canadian data residency.
Using AI inside your app#
If your app itself calls AI models, the AI gateway gives it OpenAI-, Anthropic- and Google-model access with no API keys of its own, paid from your organization's prepaid AI credits. Those models run at providers in the United States, so external AI is off until an owner or admin of your organization acknowledges that.