Power Platform ALM MCP connector for solution and pipeline management

June 18, 2026

Solution management in Power Platform still involves too many portal clicks. Export a solution, switch environments, import it, publish customizations, check the pipeline status — each step is a separate browser session. This connector collapses that workflow into operations that Copilot Studio agents and Power Automate flows can call directly.

The connector is dual-mode: a single POST /mcp endpoint for Copilot Studio agents, plus typed Power Automate operations with full schemas, dynamic dropdowns, and IntelliSense. It talks to the Dataverse Web API for solution and pipeline operations, and the Power Platform Admin API for environment discovery.

You can find the complete code in my SharingIsCaring repository.

What it does

The connector covers two ALM domains: solution lifecycle and deployment pipelines.

Solution management

Operation MCP tool API call
List Solutions alm_list_solutions GET /api/data/v9.2/solutions
Get Solution alm_get_solution GET /api/data/v9.2/solutions?$filter=uniquename eq '{name}'
Export Solution alm_export_solution POST /api/data/v9.2/ExportSolution
Import Solution alm_import_solution POST /api/data/v9.2/ImportSolution
Publish Customizations alm_publish_customizations POST /api/data/v9.2/PublishAllXml
Delete Solution alm_delete_solution DELETE /api/data/v9.2/solutions({id})
Set Solution Version alm_set_solution_version PATCH /api/data/v9.2/solutions({id})
Clone as Patch alm_clone_as_patch POST /api/data/v9.2/CloneAsPatch
Clone as Solution alm_clone_as_solution POST /api/data/v9.2/CloneAsSolution
Delete and Promote alm_delete_and_promote POST /api/data/v9.2/DeleteAndPromote

The import operation supports holding solutions for staged upgrades. Export a solution as managed, import it as a holding solution in the target environment, validate, then call Delete and Promote to complete the upgrade. That three-step pattern is the safest way to push managed solutions without downtime, and now it’s automatable in a single flow.

Pipeline management

Operation MCP tool Entity
List Pipelines alm_list_pipelines deploymentpipelines
List Pipeline Stages alm_list_pipeline_stages deploymentstages
Deploy to Pipeline alm_deploy_to_pipeline deploymentstageruns
Get Deployment Status alm_get_deployment_status deploymentstageruns
List Deployment History alm_list_deployment_history deploymentstageruns

If your org already uses Power Platform pipelines, these operations let you trigger deployments and poll status from a flow or agent instead of the maker portal.

Architecture

Authentication

The connector uses OAuth 2.0 against https://api.powerplatform.com via the aad identity provider. It calls the Power Platform Admin API to list environments and resolve Dataverse org URLs, then forwards the auth token to individual Dataverse Web APIs via Context.SendAsync.

Dynamic dropdowns

Three internal operations populate dropdowns in Power Automate:

  • GetEnvironmentDropdown — filters to Dataverse-provisioned environments
  • GetSolutionDropdown — filters to unmanaged solutions in the selected environment
  • GetPipelineDropdown — populates pipelines for the selected host environment

These eliminate the need to look up environment IDs or solution GUIDs manually.

Dual-mode pattern

The connector follows the standard dual-mode layout:

  • basePath: "/" with the MCP endpoint at path: "/mcp"
  • x-ms-agentic-protocol: "mcp-streamable-1.0" on the MCP operation
  • All operations routed through script.csx via scriptOperations in apiProperties.json

Example Copilot Studio prompts

With the MCP server attached as an action, a Copilot Studio agent can handle ALM tasks conversationally:

  • “List all unmanaged solutions in the dev environment”
  • “Export the Contoso CRM solution as managed”
  • “Import the solution into production as a holding solution”
  • “What’s the status of the last pipeline deployment?”
  • “Set the solution version to 2.1.0.0 and export it”

Configuration

  1. Create an Azure AD app registration (multi-tenant) with delegated permissions for Power Platform API:
    • EnvironmentManagement.Environments.Read
    • EnvironmentManagement.Settings.Read
    • EnvironmentManagement.Settings.ReadWrite
    • AppManagement.ApplicationPackages.Read
    • AppManagement.ApplicationPackages.Install

  2. Add redirect URI: https://global.consent.azure-apim.net/redirect

  3. Create a service principal and grant admin consent:
az ad sp create --id <appId>
az ad app permission grant --id <appId> --api 8578e004-a5c6-46e7-913e-12f58912df43 --scope "EnvironmentManagement.Environments.Read EnvironmentManagement.Settings.Read EnvironmentManagement.Settings.ReadWrite AppManagement.ApplicationPackages.Read AppManagement.ApplicationPackages.Install"

  1. Replace [INSERT_YOUR_CLIENT_ID] in apiProperties.json with your app’s client ID

  2. Optionally replace [INSERT_YOUR_APP_INSIGHTS_CONNECTION_STRING] in script.csx for telemetry

  3. Deploy with PAC CLI:

pac connector create --api-definition-file apiDefinition.swagger.json --api-properties-file apiProperties.json --script-file script.csx

Known limitations

  • Solution export size: Custom connectors have an approximate 100 MB request body limit. Most solutions are well under this.
  • Pipeline entity names: deploymentpipelines, deploymentstages, and deploymentstageruns are the expected Dataverse entity logical names. Verify against your environment’s metadata if pipelines aren’t returning results.
  • Token forwarding: The api.powerplatform.com token is forwarded to Dataverse org URLs. If you see 401 errors during testing, the auth resource may need to change to a per-environment Dataverse org URL.

Resources

results matching ""

    No results matching ""