Skip to content
Integrations

Integrations

Connect DevGhouse to the tools you already use, and let an AI assistant like Claude work inside your account through MCP.

Integrations

DevGhouse does not ask you to move your work into a new tool. Instead it connects to the tools you already have, so tasks land in Todoist, code ships to GitHub, emails go through Gmail or Outlook, and your workspace still knows about all of it.

On top of that, DevGhouse speaks MCP (Model Context Protocol). If you use Claude, Claude Code, Cursor, or another MCP-compatible assistant, you can connect it to your DevGhouse workspace with one token. Once connected, that assistant can read and act on your projects, tasks, emails, and every connected service, the same way you would from the app.


What you can do

FeatureWhat it does
Connect an AI assistant (MCP)Get a personal token and add DevGhouse as an MCP server to Claude.ai, Claude Code, Cursor, or Windsurf. Your assistant can then manage tasks, send emails, browse code, and more, in plain conversation.
AI agents and modelsChoose which AI runs your chats and automations, and link your own Claude, Codex, or Kimi subscription so runs use your plan instead of DevGhouse credits.
GitHub and VercelBrowse and edit real repositories, open pull requests, and deploy, all from inside DevGhouse.
FigmaPull frames, components, and comments from a Figma file into your workspace.
Todoist, Trello, NotionSync tasks, boards, and notes so your planning lives where you already keep it.
Gmail, Outlook, TelegramRead and send messages, and let an assistant draft replies for the ones that need one.
CalendlyShow a "Book a meeting" link on your public profile.
X (Twitter)Compose, schedule, and publish posts, and track mentions and messages.
Telegram operatorTurn your own Telegram bot into a personal remote control for your agents.

Who this is for

Anyone with a DevGhouse account. Connections are personal: what you connect under your own account is yours, and a teammate in the same workspace connects their own copy separately. A few connections (Todoist, GitHub, and others reached through a module) can be shared with someone else on your team so they can use your connection with their own agents, without you handing over your login.

How it connects to your workspace

Every connection lives on your account or on the shared workspace you are currently in, whichever one you connected it under. Switch context with the picker in the top bar and you are working with that context's own connections and its own credits. A DevGhouse module (Coding, Planning, Communication, Social) only lights up its related connections once that module is turned on for the context you are in; turning a module on or off is done from your workspace home under Modules.

Agents and routines

Every connection you make by hand is also a tool an AI agent can call for you, on your behalf, once you tell it to. The MCP Server page walks through connecting an assistant. Once it is connected, here is roughly where each integration's tools live for an agent to use:

  • Coding: github_list_repositories, github_create_or_update_file, github_create_issue, vercel_create_deployment, and the rest of the GitHub and Vercel surface.
  • Assets: figma_get_file, figma_export_images, figma_get_comments.
  • Planning: todoist_create_task, todoist_list_projects, trello_create_card, notion_create_page, calendly_list_scheduled_events.
  • Communication: gmail_send_email, outlook_list_emails, telegram_send_message, telegram_bot_send_message.
  • Social: twitter_post_tweet, twitter_get_mentions, twitter_send_dm.
  • Core (always available, no module needed): get_api_tokens, get_default_ai_models, set_default_ai_models.

An agent only sees the tools for a service you have actually connected. Connect Gmail yourself first, and then your assistant can read and send mail through it; it never sees a service you have not turned on.