Set up Logic Foundry

Logic Foundry is installed and hosted by your administrator, once per Dataverse environment. As a designer, you then add it to your AI client with the address your administrator gives you and sign in with your own account. This page covers both sides at the level you need to plan and request a setup.

For administrators

To deploy Logic Foundry, see Deploy Logic Foundry and the Logic Agents.

Before you begin

Requirement	Notes
Logic Composer	Installed in the environment. It is the basis for most tools.
Process Orchestrator	Optional. Needed only for the business process tools.
An MCP-capable AI client	For example GitHub Copilot, Claude or Visual Studio Code.
Access granted by your administrator	Your administrator decides who can connect to Logic Foundry.
A Dataverse user	You work with your own security roles. An assistant can’t do anything you can’t do yourself.

What your administrator decides

When your administrator installs Logic Foundry, they make three decisions that affect how you work.

Decision	What it means for you	Recommendation
Which environment	Each Logic Foundry installation connects to exactly one Dataverse environment. The assistant sees only that environment.	One installation per environment: development, test, production.
How people sign in	With per-user sign-in, you sign in with your organization account and every tool call runs as you. A shared service account makes every call run with one account’s permissions, whoever is asking.	Use per-user sign-in for any shared installation. Keep a shared account for a personal sandbox only.
Who can connect	Only people granted access can add Logic Foundry to their AI client.	Grant access to designers and consultants who build logic.

Caution

Connect AI clients to a development environment for building. In test and production environments, projects normally belong to managed solutions, so Logic Foundry refuses changes there anyway, but connecting to the right environment from the start avoids surprises when you run flows.

Add Logic Foundry to your AI client

1. Ask your administrator for the Logic Foundry address for the environment you want to work in.
2. In your AI client, add a new MCP server of type HTTP (some clients call it remote or streamable HTTP) and paste the address. Give it a recognizable name that includes the environment, such as logic-foundry-dev.
3. Save the configuration. When the client connects for the first time, sign in with your organization account and consent if asked.
4. Check that the tools are listed. Most clients show the number of tools or let you browse them.
5. Test the connection with a read-only question, for example: List my Logic Fabric projects.

Each client stores MCP servers slightly differently. In Visual Studio Code, for example, you add an entry to the workspace’s MCP configuration:

{
  "servers": {
    "logic-foundry-dev": {
      "type": "http",
      "url": "https://<address from your administrator>"
    }
  }
}

Check your AI client’s documentation for the exact place and format.

Work with more than one environment

Add a separate entry for each environment, each with its own address and a name that says which environment it is. Turn on only the entry you need for the task at hand, so the assistant can’t mix up environments.

Troubleshooting

Symptom	What to check
The client can’t connect or keeps asking you to sign in	The address is correct and your administrator has granted you access.
A project or artifact is missing	You have read access to it in Dataverse. The assistant sees only what you can see.
Creating or updating fails with a managed-solution error	The project belongs to a managed solution in this environment. Make the change in development.
Business process tools fail	Process Orchestrator is installed in the environment.

Related

• Logic Foundry tools
• Working with AI
• Set up Logic Conductor