Documentation

Power Platform Skills

Everything you need to install, configure, and use the PP Skills MCP server with your AI agent.

Overview

Power Platform Skills is an MCP (Model Context Protocol) server that gives any AI agent expert-level Microsoft Power Platform development knowledge. One install command, and your AI assistant knows how to build Dataverse schemas, Power Apps, Power Automate flows, PCF controls, plugins, Power BI reports, and more.

Endpoint

api.ppskills.dev/mcp

Protocol

MCP (Streamable HTTP)

Skills

179 total

Playbooks

2 end-to-end

Installation

Free tier — no API key needed

claude mcp add power-platform-skills --transport http https://api.ppskills.dev/mcp

Restart Claude Code after installing.

With Pro API key

claude mcp add power-platform-skills --transport http https://api.ppskills.dev/mcp --header "Authorization: Bearer sk_pro_YOUR_KEY_HERE"

Any MCP client: Streamable HTTP at https://api.ppskills.dev/mcp. Add Authorization: Bearer sk_pro_... header for pro access.

MCP Tools

The server exposes 4 tools. Your AI calls these automatically when you ask Power Platform questions.

list_skills

No arguments

Returns a structured catalog of everything available — playbooks and skills grouped by category. The AI should call this first when starting any Power Platform task.

User: "What Power Platform skills do you have?"
AI calls: list_skills()
→ Returns catalog with playbooks + all skills by category

get_skill

skill_id (string, required)

Fetches a specific skill by its exact ID. Returns the full expert knowledge — API endpoints, JSON schemas, code patterns, best practices.

User: "How do I create a Dataverse table?"
AI calls: get_skill({ skill_id: "dataverse-web-api--tables-entities" })
→ Returns full guide with exact API endpoints, JSON payloads, and gotchas

get_playbook

playbook_id (string, required)

Returns a step-by-step playbook for building something end-to-end. Each step tells the AI exactly which skill to load and what to do with it.

User: "Build me a Power BI report from this data"
AI calls: get_playbook({ playbook_id: "build-power-bi-report" })
→ Returns 8-phase guide, AI follows each step calling get_skill() as directed

get_playbook_phase

playbook_id (string, required), phase (string, required)

Loads a specific phase from a playbook with all its skill content bundled in one response. Use this instead of calling get_skill multiple times per phase — ensures the AI has every resource it needs.

AI calls: get_playbook_phase({ playbook_id: "build-power-bi-report", phase: "Phase 4" })
→ Returns Phase 4 instructions + full content of all 4 required skills

search_power_platform_skills

query (string, required), max_results (number, 1-5, default 3)

Keyword search across all skills. Use when exploring a topic without knowing the exact skill ID.

User: "How do adaptive cards work in Power Automate?"
AI calls: search_power_platform_skills({ query: "adaptive cards" })
→ Returns matching skill content with code examples

Recommended flow

list_skills()

See what's available

get_playbook(id)

Step-by-step guide

get_playbook_phase()

Phase + all skills

search(query)

Explore by keyword

Skills Reference

Each category has an orchestrator skill (the --skill suffix) that defines the workflow, plus resource skills with detailed reference material.

Basic Tier

Free63 files across 8 categories

Pro Tier

Pro116 additional files across 14 categories

Everything in Basic, plus:

Playbooks

Playbooks are the most powerful feature. They turn the AI into a guided builder that follows a proven step-by-step process.

Build a Power BI Report

build-power-bi-report

Takes raw data to a polished, story-driven Power BI report (PBIP format).

Phase	Name	What Happens
Phase 0	Data Discovery	Profile semantic model, discover measures, flag data issues
Phase 1	Story Finding	Explore data for insights, build narrative arc, critique until 9/10
Phase 1.5	Data Model Prep	Map insights to measures, identify model enhancements
Phase 2	Report Design	Progressive page specs with critique loops — theme, cover, pages, interactions
Phase 2.5	Architecture	Plan PBIP folder structure and build sequence
Phase 3	Specification	Exact positions, data bindings, formatting for every visual
Phase 4	Build PBIP	Assemble folder structure — scaffold, theme, pages, interactions
Phase 5	Visual QA	6-tier progressive QA from "does it open?" to "can we elevate it?"

Build a Power Platform Solution

build-dataverse-solution

Takes an idea to a fully deployed Power Platform solution.

Phase	Name	What Happens
Phase 1	Spec & Plan	11-phase spec pipeline: research, requirements, design, build plan
Phase 2	Solutions & Publisher	Multi-solution structure (CoreData, Apps, Automation, Analytics)
Phase 3	Schema — Tables & Columns	Table shells sequential, columns parallel per table
Phase 4	Schema — Relationships	Lookups, N:N, cascade behaviors
Phase 5	Security	Roles, column security, team access
Phase 6	Model-Driven App	Forms, views, business rules, app module — all parallel
Phase 7	Web Resources & Code Apps	JS scripts, ribbon buttons, HTML dashboards, embedded apps
Phase 8	Automation	Power Automate flows + C# plugins
Phase 9	Publish & Test	Publish all, API validation, browser E2E testing

Example Prompts

Copy-paste these into your AI agent to try PP Skills.

Getting Started

What Power Platform skills do you have?

Power BI Report — Full Build

I want to build a Power BI report from my sales data. The semantic model is already deployed to Fabric workspace "Sales Analytics". Start with the playbook for building a Power BI report and guide me through each phase.

Power BI Report — Quick

Create a Power BI report that shows monthly revenue trends, top 10 customers by revenue, and a regional breakdown. Use a dark theme with accent green. The semantic model has tables: Sales, Customers, Products, Calendar.

Dataverse Solution — Full Build

Create a spec for a customer complaints management system. It needs: a complaints table with status tracking, an approval flow for escalations, a model-driven app for support staff, and a dashboard showing resolution times. Follow the full Power Platform solution playbook.

Dataverse Table Creation

Create a Dataverse table called "Project" with columns: Name (text), Status (choice: Not Started/In Progress/Completed/On Hold), Start Date (date), End Date (date), Budget (currency), Owner (lookup to systemuser), and Priority (choice: Low/Medium/High/Critical). Add it to a solution called "ProjectTracker" with publisher prefix "pt".

Power Automate Flow

Build a Power Automate flow that triggers when a new record is created in the "Support Ticket" Dataverse table. If Priority is "Critical", send an approval request to the team lead with an Adaptive Card showing ticket details. If approved, update the ticket status to "Escalated" and send a Teams notification.

Model-Driven App

Build a model-driven app for the HR onboarding system. It needs: a sitemap with areas for "Onboarding" (new hire records, tasks, documents) and "Admin" (settings, templates). Create Main forms with tabs for Details, Tasks, and Documents. Add quick create forms for new hires and tasks.

PCF Control

Create a PCF control that displays a star rating (1-5 stars) for a whole number field. It should be interactive (clickable), show hover states, and work on both main forms and editable grids. Use React for the rendering.

Plugin Development

Write a C# plugin that runs on Pre-Create of the "Invoice" table. It should auto-generate a sequential invoice number in the format "INV-2026-00001" by querying the latest invoice and incrementing. Register it as a synchronous pre-operation step.

Code Review

Review the Dataverse solution in this directory. Check for: security role gaps, missing form validations, flow error handling, plugin registration issues, and any schema design problems. Use the code review skill for the audit.

Web Resources

Create a JavaScript web resource for the Contact form that: shows/hides the "Company Name" field based on the Contact Type choice, auto-populates the "Full Name" field from First + Last name on change, and validates that Email is filled when Contact Type is "Business". Register it on the Main form.

Visual QA

QA the Power BI report I just built. Open it in Desktop, go through all pages, and run the full 6-tier visual review. Score each page and give me specific fixes for anything below 9/10.

Pricing

	BasicFree	Pro$45/mo	Team$149/mo	Org$399/mo
Skills	63 (8 categories)	179 (all)	179 (all)	179 (all)
Playbooks	—	2	2	2
Requests/hour	20	1,000	5,000	100,000
API Keys	No key needed	1 key	10 keys	Unlimited
Support	—	Email	Priority	Dedicated + PO/Invoice

FWhat's in Basic (Free)?

Dataverse Web API (20 resources)
Dataverse Plugins (5 resources)
Dataverse Web Resources (9 resources)
PCF Controls (4 resources)
Power Apps Code Apps (9 resources)
Plan with Team (6 resources)
Code Review (5 resources)
Visual QA (5 resources)

PWhat does Pro add?

Playbooks — guided end-to-end builds
Power Automate (11 resources)
Power BI Report (29 resources)
Power BI Model (9 resources)
Power BI Story (16 resources)
Power Platform Spec (12 resources)
PBI Review — offline + online (22 resources)
PBI Smoke Test (4 resources)
Additional Dataverse — BPF, dashboards, schemas

API Reference

Endpoint

POST https://api.ppskills.dev/mcp

MCP (Model Context Protocol) using Streamable HTTP transport. All requests are JSON-RPC 2.0.

Authentication

FreeNo key needed

Pro+Pass API key via Authorization: Bearer sk_pro_... header

Health Check

GET https://api.ppskills.dev/health\n→ {"status":"ok","service":"ppskills-mcp"}

Rate Limits

Tier	Requests/Hour
Basic	20
Pro	1,000
Team	5,000
Org	100,000

When exceeded, returns HTTP 429 with Retry-After: 60 header.

Security & Trust

PP Skills is a read-only documentation server. It has no ability to access your environment, and it never sees your data. Here is exactly how the architecture works.

How the connection works

Your AI agent connects to our API at api.ppskills.dev/mcp over HTTPS and sends a JSON-RPC request asking for a skill or search result. We return markdown content — documentation, code patterns, and best practices. That's the entire interaction.

Your agent then uses that knowledge to act on your behalf in your own environment, using your own credentials stored on your machine. We are never part of that second step. We have no connection to your Dataverse instance, your Azure tenant, your DevOps organisation, or any other service you use.

Think of it like a reference book — your agent reads it, then goes and does the work. The book never touches the project.

What we store

We store the absolute minimum required to run the service. Nothing else is logged, cached, or retained.

Data	Purpose	Stored in
Email address	Account identification and billing	Stripe
API key hash	Request authentication	Cloudflare KV
Hourly request count	Rate limiting	Cloudflare KV (auto-expires)

We do not store your search queries, tool call history, request bodies, IP addresses, or any content your agent generates.

What we never do

Execute code in your environment
The MCP server is read-only. It returns text. It cannot run commands, call APIs on your behalf, or interact with any external service.
Connect to your Dataverse, Azure, or DevOps
We have no access to your tenant, subscriptions, or service principals. Your agent uses its own credentials locally after reading our skills.
See your credentials or tokens
Your environment credentials (service principals, connection strings, OAuth tokens) never leave your machine. They are used by your local AI agent, not by our server.
Log or retain your queries
Request bodies are not logged. We track a numeric request count for rate limiting — that count auto-expires hourly and contains no content.
Access or store customer data
We never see the data your agent works with. If your agent builds a Dataverse table or Power BI report, that happens entirely in your environment.
Train models on your interactions
We do not use any request data for training, fine-tuning, or analytics. We are a documentation server, not an AI provider.

Infrastructure

Layer	Provider	Details
API server	Cloudflare Workers	Edge-deployed, SOC 2 Type II, ISO 27001 certified
Data storage	Cloudflare KV	Encrypted at rest, API keys and rate limit counters only
Payments	Stripe	PCI DSS Level 1 — we never see or store card numbers
Transport	HTTPS / TLS 1.3	All traffic encrypted in transit, no plaintext endpoints
Source code	GitHub (public)	Server code is open-source — audit the full request path yourself

API key security

Pro, Team, and Org API keys are prefixed with sk_pro_, sk_team_, or sk_org_ so they are easy to identify in your configuration files. Keys are sent via the standard Authorization: Bearer header over HTTPS.

If a key is compromised, you can rotate it immediately from your Stripe billing portal. The old key is revoked and a new one is issued. Your API key only grants access to PP Skills content — it cannot be used to access any other service.

FAQ

Do I need to know which skill to ask for?

No. Just ask your question naturally. The AI will call list_skills or search_power_platform_skills to find the right skill automatically.

What AI clients work with this?

Any MCP-compatible client: Claude Code, Claude Desktop, VS Code (Copilot, Continue), Cursor, Windsurf, GitHub Copilot, Copilot Studio, Azure AI Foundry, and more.

How current is the knowledge?

Skills are updated weekly with the latest API patterns, schema changes, and best practices. More current than any AI's training data.

Can I use it for real production work?

Yes. The skills contain production-grade patterns with exact API endpoints, JSON schemas, error handling, and security considerations. The playbooks follow battle-tested build processes.

What's the difference between skills and playbooks?

Skills are reference material (API docs, code patterns, best practices). Playbooks are step-by-step guides that tell the AI which skills to load and what to do at each phase of a big build.

I'm on the free tier — what can I do?

Build Dataverse schemas (tables, columns, relationships, forms, views), write plugins, create PCF controls, build code-first Power Apps, and more. The free tier covers the core data layer and app building.

What does Pro add?

Power Automate flows, Power BI reports (the full story-driven pipeline), the spec pipeline for generating solution designs, advanced web resources, and the playbooks for end-to-end guided builds.

Ready to get started?

Install in 30 seconds. Free to start. No credit card required.

Get Started Free View Pricing