What is MCP?
MCP — the Model Context Protocol — is an open standard that gives an AI assistant a structured way to use other software. Before it, wiring an assistant to a tool meant a custom integration for every pairing, built and maintained by hand. MCP replaces all of that with one protocol: a tool publishes a connector that speaks it, and any assistant that speaks it — a client — can use that connector.
Dalus publishes such a connector. Connect it once, and the assistant you already use can query your system model, run the analyses inside it, and stage changes back — requirements, architecture, budgets, verification, all of it. Every workflow in the library is simply a prompt written against that connection. The protocol itself is not ours: it is an open industry standard, which is why one library of prompts can serve Claude, ChatGPT, Cursor and VS Code alike.
Claude
ChatGPT
Cursor
VS CodeThe chat you already work in. Any client that speaks the protocol connects the same way.
Speaks the protocol for your workspace. It carries your login and your permissions — the assistant sees what you can see, and nothing else.
Requirements, architecture, interfaces, budgets, hazards and verification — the live model, not an export of it.
Connect once
Add the Dalus connector to your client's settings and sign in with your Dalus account. There is no server to run, nothing to install, and nothing to build.
Ask in plain language
Paste a prompt from the workflow library, or just describe what you need. The assistant reads the model through the connection instead of through exports and copy and paste.
Approve anything that writes
Reads simply happen. Anything that would change the model is shown to you as a full change set first, and nothing lands until you say yes.
Which requirements in the propulsion model have no passing verification?
Reads the requirement set and its verification items from the live model, then answers with the six requirement IDs and what each one is missing.
A standard, not an integration you build
MCP is published as an open protocol. Dalus runs a connector that speaks it, and any assistant that speaks it can use the connector. Nothing is written per team, and no plugin is installed.
The work happens where you already write
You stay in Claude, ChatGPT, Cursor, VS Code or another MCP client. The prompt goes in the chat, and the assistant reaches the model through the connection.
It runs when you ask it to
There is no background sync and no daemon watching your model. A workflow runs because somebody started it, which is also why every run has an obvious author.
It works as you, with your permissions
The connection carries your access, so it reads what you can read and no more. That is also why so many of the library's prompts end by asking for approval: writing is a decision, and the decision stays yours.
See it in the workflows
Every workflow in the library shows the prompt it runs, in full, before you run it — which makes them the fastest way to understand what a connected assistant can actually do. For the systems the connector reaches, see Dalus integrations.
Browse the workflow library