MCP Power Automate

Local MCP server and Chromium extension for AI-operated Microsoft Power Automate flows.

The extension captures the browser session, current flow, snapshots, and token candidates automatically. The MCP server exposes one clean v1 tool surface so the AI can inspect, edit, validate, run, review, and revert flows without asking the user to click extension buttons just to make the connection work.

No Microsoft Entra ID app registration, admin consent, or custom enterprise app setup is required.
The MCP uses your existing logged-in Chromium session.

What v1 Changes

• One command registry powers both MCP tools and HTTP bridge routes.
• The local bridge is authoritative; reused MCP instances proxy commands to the bridge owner.
• Flow targeting is automatic-first through captured tabs, explicit connect_flow, catalog data, and snapshots.
• The extension is status and diagnostics only. It no longer requires manual refresh/sync buttons for normal operation.
• npm ships the built server and built extension together.

Install

Register the MCP:

codex mcp add power-automate-local -- npx -y @kaael1/mcp-power-automate

Find the packaged extension path:

npx -y @kaael1/mcp-power-automate extension-path

Load that folder in Chromium:

1. Open chrome://extensions or edge://extensions
2. Enable Developer Mode
3. Choose Load unpacked
4. Select the path printed by extension-path

Open or focus any Power Automate flow page. The extension captures the context automatically.

Check readiness:

npx -y @kaael1/mcp-power-automate doctor

Recommended AI Workflow

Ask your MCP client to:

1. doctor
2. get_context
3. connect_flow
4. get_flow
5. preview_flow_update
6. validate_flow
7. apply_flow_update
8. get_last_update

For run inspection and manual/request trigger tests, use list_runs, get_latest_run, get_run, get_run_actions, wait_for_run, get_trigger_callback_url, and invoke_trigger.

For Dataverse solution inspection, use list_solutions, list_solution_components, and list_environment_variables. These tools are read-only in this release.

Public v1 Tools

• get_context
• doctor
• connect_flow
• list_flows
• list_solutions
• list_solution_components
• list_environment_variables
• get_flow
• preview_flow_update
• validate_flow
• apply_flow_update
• get_last_update
• revert_last_update
• list_runs
• get_latest_run
• get_run
• get_run_actions
• wait_for_run
• get_trigger_callback_url
• invoke_trigger
• create_flow
• clone_flow

Solutions And Environment Variables

The extension can capture Power Platform/BAP and Dataverse-audience tokens from the logged-in browser session. When those tokens are present, get_context and /health expose canManageSolutions and the MCP can inspect Dataverse solutions for the current or provided environment.

The first solution-oriented surface is intentionally read-only:

• list_solutions lists visible unmanaged solutions by default.
• list_solution_components lists components for a solution unique name, with optional enrichment for cloud flows and environment variables.
• list_environment_variables lists environment variable definitions and current values, optionally scoped to one solution.

Write operations such as creating solutions, setting environment variable values, publishing customizations, deleting components, or deleting flows are not exposed in this phase.

HTTP Bridge

The bridge listens on 127.0.0.1:17373.

• GET /health is kept for simple probes.
• GET /v1/health returns bridge identity and readiness.
• GET /v1/context returns the same context used by the MCP.
• GET /v1/commands lists the public v1 command surface.
• POST /v1/commands/:name runs any public v1 command with a JSON body.

Only the process that owns the bridge port executes stateful work. Other MCP instances reuse the healthy bridge and proxy commands to it.

Safety Model

• Use preview_flow_update before saves.
• Use validate_flow before and after meaningful edits when available.
• Use get_last_update to review the persisted diff.
• Use revert_last_update if the saved result is wrong.
• Prefer test or staging flows before production flows.

If Power Automate rejects a save because of a connection permission problem, the MCP reports CONNECTION_AUTHORIZATION_FAILED and waits for the user to fix that connection in Power Automate. If the service rejects a field such as retryPolicy, the MCP reports SCHEMA_VALIDATION_FAILED with the rejected member so the AI can correct the candidate flow instead of guessing.

Development

npm install
npm run typecheck
npm run lint
npm run test
npm run build
npm run pack:dry-run

For a local clone:

npm run build
codex mcp add power-automate-local -- node C:\path\to\mcp-power-automate\dist\server\index.js
node C:\path\to\mcp-power-automate\dist\server\index.js extension-path

Runtime state lives in data/ and must not be committed.

Package Links

• GitHub: https://github.com/kaael1/mcp-power-automate
• npm: https://www.npmjs.com/package/@kaael1/mcp-power-automate
• MCP Registry: https://registry.modelcontextprotocol.io/v0/servers?search=io.github.kaael1/mcp-power-automate

License

MIT. See LICENSE.