RimbraDocs

Rimbra MCP server

Connect Claude, ChatGPT, Cursor or any MCP client to your workspace so it can deploy, read logs, run databases and send messages.

What an agent can do

Rimbra runs an MCP server (Model Context Protocol, the standard most AI agents speak) at https://app.rimbra.app/mcp. A connected agent works in one workspace with the same plan limits and team roles as a person. It can never pay for anything, buy domains, or change billing or the team.

AreaTools
Workspacewhoami, list_team, contact_support
Projectslist_projects, get_project, create_project, update_project_settings, list_github_repos
Deploydeploy_project, get_deployment, cancel_deployment, get_runtime_logs, rollback_project
Variableslist_env_vars, set_env_var
Databaseslist_databases, create_database, connect_database, query_database, list_backups, backup_database, restore_backup
Domainslist_domains, add_domain, search_domains
Emaillist_email_domains, add_email_domain, send_email, list_emails
WhatsApplist_whatsapp_numbers, send_whatsapp

Claude Code

Install the Rimbra plugin. It adds the connection and a skill that teaches Claude how Rimbra works.

code
/plugin marketplace add AdejareAkolawole/rimbra-agents
/plugin install rimbra@rimbra

Claude.ai and Claude Desktop

  1. 1Open Settings → Connectors → Add custom connector.
  2. 2Name it Rimbra and paste https://app.rimbra.app/mcp.
  3. 3Choose Connect, sign in to Rimbra, pick the workspace and access, and choose Allow.

ChatGPT

  1. 1Open Settings → Apps & Connectors, and turn on developer mode under Advanced.
  2. 2Create a connector with the address https://app.rimbra.app/mcp and OAuth sign-in.
  3. 3ChatGPT sends you to Rimbra to approve it.

Cursor and other clients

Use Add to Cursor in Rimbra → MCP, or add this to ~/.cursor/mcp.json. Any client that speaks Streamable HTTP works the same way.

json
{ "mcpServers": { "rimbra": { "url": "https://app.rimbra.app/mcp" } } }

Access and keys

  • Signing in (OAuth) is best: nothing to copy, and access tokens last an hour and refresh themselves.
  • Keys are for agents that can’t sign in, like a terminal or CI. They start with rmb_ and are shown once.
  • Developer access can deploy and change things. Viewer access only reads.
  • An agent never gets more access than the person who connected it. A team Viewer can only connect Viewer agents.
  • See and disconnect every agent and key in Rimbra → MCP.

For agents that read docs

A short plain-text guide to Rimbra for language models lives at /llms.txt.