Setup logo

Setup

Organization
microsoft
setup

Set up Power Automate CLI prerequisites. Use when the user is new, something isn't working, or they need help getting started.

Overview

Publishermicrosoft
Repositorypower-platform-skills
Skill namesetup
Stars
993
Forks
204
Bundled files
Instructions only
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by microsoft on GitHub. Read the source before you install it.

Installation

Install the Setup AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/microsoft/power-platform-skills/tree/main/plugins/power-automate/skills/setup
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/microsoft/power-platform-skills.git /tmp/power-platform-skills
mkdir -p .claude/skills
cp -r /tmp/power-platform-skills/plugins/power-automate/skills/setup .claude/skills/setup
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Setup in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Setup on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Setup is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

First-Time Setup Guide

You are helping a non-technical user get the Power Automate plugin working for the first time. Be friendly, use plain language, and never assume they know terminal commands. Walk them through each step one at a time.

Step 1: Check Node.js

Run silently:

bash
node --version 2>&1
  • If it works (prints something like v18.x.x or higher): Tell them "Node.js is installed" and move on.
  • If it fails or version is below 18: Tell them they need Node.js 18 or newer. Ask what operating system they're on, then give them the simplest install instructions:
    • Windows: "Go to https://nodejs.org, download the LTS version, and run the installer. Click Next through everything."
    • Mac: "Open Terminal and run: brew install node" (or direct them to nodejs.org)
    • After they've installed it, re-check with node --version.

Step 2: Check Azure CLI

Run silently:

bash
az --version 2>&1
  • If it works: Tell them "Azure CLI is installed" and move on.
  • If it fails: Tell them they need the Azure CLI. Ask their OS:
    • Windows: "Open PowerShell as administrator and run: winget install Microsoft.AzureCLI" or direct them to https://aka.ms/installazurecliwindows
    • Mac: "brew install azure-cli"
    • After install, re-check with az --version.

Step 3: Azure Login

Check if they're already logged in:

bash
az account show --output json 2>&1
  • If it works (shows account info): Tell them who they're logged in as (show the user.name field) and ask if that's the right account.
  • If it fails: Tell them "Let's sign you into Azure." Then run:
    bash
    az login
    This will open their browser. Tell them: "A browser window should open. Sign in with your work account — the one you use for Power Automate." After login completes, confirm it worked by running az account show again.

Verify token access — this catches permission issues early.

First find out which Azure cloud they're on, because the Power Automate resource URL differs per cloud and the commercial one cannot be assumed (GCC High / DoD tenants will fail against it):

bash
az cloud show --query name -o tsv
az cloud showPower Automate resource
AzureCloud (commercial)https://service.flow.microsoft.com
AzureCloud + GCC tenanthttps://gov.service.flow.microsoft.us
AzureUSGovernment (GCC High)https://high.service.flow.microsoft.us
AzureUSGovernment (DoD)Operator-verified PA_FLOW_RESOURCE value required; no built-in audience

az cloud show cannot distinguish commercial from GCC, or GCC High from DoD — for those, set PA_CLOUD=gcc / PA_CLOUD=dod explicitly.

DoD endpoint hosts use appsplatform.us, but that does not establish the token audience (App ID URI). With PA_CLOUD=dod, FlowAgent requires an explicit, operator-verified PA_FLOW_RESOURCE and fails configuration otherwise. Do not copy a guessed audience from a hostname; obtain the correct value for the tenant.

Then request a token for the matching resource, e.g. for commercial:

bash
az account get-access-token --resource https://service.flow.microsoft.com --output json 2>&1

FlowAgent itself auto-detects the cloud the same way; you can override the detection with PA_CLOUD=commercial|gcc|gcchigh|dod.

  • If it works: Move on.
  • If it fails with "AADSTS": The user's account may not have Power Automate access. Tell them: "Your Azure account doesn't seem to have access to Power Automate. Check with your IT admin that you have a Power Automate license."
  • If it fails with other errors: Show the error and suggest they contact IT support.
  • If they're on a sovereign cloud and connection-management commands fail: keep the original error and inspect its code. AADSTS650057 can indicate an invalid resource or missing app authorization; AADSTS65001 can require admin consent. An app-not-found or consent failure may require an appropriately authorized public-client registration and PA_CLIENT_ID=<app-id>. Network, tenant-selection and expired-session errors need different remediation. Do not infer missing preauthorization from the cloud name or classify every other code as unrelated.

Signing in to a specific account

Connection management authenticates separately from az — it uses its own MSAL session with its own on-disk token cache, so az login / az account set do not switch the account it uses.

Three tools cover this:

ToolUse it to
list_accountsSee the cached-account inventory, effectiveConnectivityIdentity, and nextSignIn settings. Inactive tenant caches do not count as identity mismatches. Acquires no token.
switch_accountSave a preferred username, or omit it to clear the preference. Check preferencePersisted and nextSignIn: environment overrides still apply.
whoami / doctorCompare the effective Connectivity username and tenant with Azure CLI, including different users in the same tenant.

When a connection tool fails with ServiceToServiceEnvironmentNotFound, check identity before assuming the environment is missing. list_accounts distinguishes the effective account from historical cached accounts. A different username or tenant on the effective account indicates a mismatch; an unused tenant's cache alone does not. switch_account clears the cached sign-in and records the preferred username for reauthentication. It does not prove that the subsequent sign-in succeeded or that environment permissions are correct.

switch_account does not change the Azure CLI identity. az is yours to set; where the two disagree, the tool says so rather than silently re-pointing one. reconnect keeps a recorded preference — it drops credentials, and the preference is a stated intent rather than a credential. It waits for pending token acquisition and cache cleanup, then clears in-memory tokens for all Azure CLI resources as well as Connectivity. A cleanup failure is an error, not a completed switch: fix the cache access problem and retry. Do not continue under the assumption that credentials changed when reset failed.

On interactive sign-in FlowAgent forces the account picker by default, so the browser's currently-signed-in account is never used silently. Two overrides:

VariableEffect
PA_LOGIN_HINT=<upn>Pre-select that account. Suppresses the picker, since the account is already targeted.
PA_NO_ACCOUNT_PICKER=1Restore plain browser SSO. For single-account users who don't want the extra click.

Precedence, most specific first: PA_LOGIN_HINT, then a switch_account preference, then PA_NO_ACCOUNT_PICKER, then the picker. For example, switch_account with username B still targets A if PA_LOGIN_HINT=A is set. Omitting the username clears only the stored preference; it does not override PA_LOGIN_HINT or PA_NO_ACCOUNT_PICKER. The response reports the effective settings instead of promising a picker in those cases.

Step 4: Check the FlowAgent tools are wired

The plugin talks to Power Automate through the FlowAgent MCP server, which is registered as flowagent in the plugin's .mcp.json and started automatically. .mcp.json loads the bundled server/mcp.mjs through a small Node bootstrap that resolves the plugin's installation directory (PLUGIN_ROOT, else CLAUDE_PLUGIN_ROOT, else the current directory) and prints an actionable error if the bundle can't be found.

  • If flowagent-* / mcp__flowagent__* tools appear in your tool list: tell them "The Power Automate tools are connected" and move on.

  • If they're missing: the MCP server isn't registered. Fix it automatically:

    1. Locate the installed plugin's MCP bundle. This only matches a bundle inside a power-automate plugin directory, so it can't pick up another plugin's MCP server:

      bash
      node -e "const fs=require('fs'),p=require('path'),d=p.join(process.env.HOME||process.env.USERPROFILE,'.copilot','installed-plugins');const find=(dir)=>{let out=[];for(const e of fs.readdirSync(dir,{withFileTypes:true})){const f=p.join(dir,e.name);if(e.isDirectory()){try{out=out.concat(find(f))}catch{}}else if(e.name==='mcp.mjs'&&p.basename(p.dirname(dir))==='power-automate'){out.push(dir)}}return out};try{const hits=find(d);if(hits.length===1)console.log(JSON.stringify({found:true,serverDir:hits[0],mcpMjs:p.join(hits[0],'mcp.mjs')}));else if(hits.length>1)console.log(JSON.stringify({found:false,reason:'multiple power-automate bundles',candidates:hits}));else console.log(JSON.stringify({found:false}))}catch(e){console.log(JSON.stringify({found:false,error:e.message}))}"

      If it reports multiple power-automate bundles, show the candidates and ask the user which one to register rather than guessing.

    2. If exactly one was found, read ~/.copilot/mcp-config.json, add the flowagent MCP entry, and write it back:

      bash
      node -e "const fs=require('fs'),p=require('path');const home=process.env.HOME||process.env.USERPROFILE;const cfgPath=p.join(home,'.copilot','mcp-config.json');let cfg;try{cfg=JSON.parse(fs.readFileSync(cfgPath,'utf8'))}catch{cfg={mcpServers:{}}};if(!cfg.mcpServers)cfg.mcpServers={};if(cfg.mcpServers.flowagent){console.log('already registered');process.exit(0)}const d=p.join(home,'.copilot','installed-plugins');const find=(dir)=>{let out=[];for(const e of fs.readdirSync(dir,{withFileTypes:true})){const f=p.join(dir,e.name);if(e.isDirectory()){try{out=out.concat(find(f))}catch{}}else if(e.name==='mcp.mjs'&&p.basename(p.dirname(dir))==='power-automate'){out.push(dir)}}return out};const hits=find(d);if(hits.length!==1){console.log(hits.length?'ambiguous: '+JSON.stringify(hits):'mcp.mjs not found');process.exit(1)}const mcpPath=p.join(hits[0],'mcp.mjs');cfg.mcpServers.flowagent={command:'node',args:[mcpPath]};fs.writeFileSync(cfgPath,JSON.stringify(cfg,null,2)+'\n');console.log('registered flowagent MCP at '+mcpPath)"
    3. Tell the user to restart the agent (Copilot CLI: /restart, Claude Code: restart the process). After restart, flowagent-* tools should appear.

    4. If not found (plugin not installed at all): tell them to install the plugin first:

      /plugin marketplace add microsoft/power-platform-skills

      Then select power-automate and run /setup again.

Step 5: Smoke Test

Verify everything works end-to-end by listing the user's environments:

  • Preferred: call the list_environments tool.

  • If MCP tools aren't available: run node <path-to-plugin>/server/mcp.mjs to confirm the bundled MCP server starts cleanly, then fix the plugin install or .mcp.json wiring before retrying.

  • If it returns environments: Success! Tell them:

    • "Everything is working! Here are your Power Automate environments:"
    • Show the environments in a simple table (name, location).
    • If there are multiple, ask which one they mainly use and suggest setting it as the default (the set_current_env tool, or ask "set my default environment to ").
    • Tell them about the available skills:
      • /browse-flows — Browse your flows
      • /create-flow — Create a new flow
      • /debug-flow — Fix a broken flow
  • If it fails: Check the error. Common issues:

    • Auth error → go back to Step 3
    • Tools not found → go back to Step 4
    • Network error → ask if they're behind a corporate proxy/VPN

Tone Guidelines

  • Use "we" language: "Let's check if Node.js is installed"
  • Celebrate small wins: "Great, Node.js is ready!"
  • Don't dump all steps at once — do one at a time and confirm before moving on
  • If something fails, don't panic — explain what went wrong in plain English and what to do
  • Never show raw JSON errors to the user without explaining what they mean

Frequently asked questions

What does the Setup AI skill do?

Set up Power Automate CLI prerequisites. Use when the user is new, something isn't working, or they need help getting started.

Why use Setup on TypingMind?

Because you install it once and use it with any model. Setup is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Setup in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/microsoft/power-platform-skills/tree/main/plugins/power-automate/skills/setup. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Setup?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Setup?

As many as you like. As long as a model supports skills, you can use Setup with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Setup AI skill free?

Yes. It is published on GitHub by microsoft under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇