Power Automate Mcp

Power Automate via FlowStudio MCP — Foundation

This 技能 is the plumbing layer. It gives an AI 代理 a reliable way to talk to a FlowStudio MCP server, discover what 工具s are avAIlable, and handle the 响应s 清理ly. The actual 工作流 narratives live in four specialized 技能s that all build on this one.

Real 调试ging examples: Expression error in child flow | Data entry, not a flow bug | Null value crashes child flow

Requires: A FlowStudio MCP subscription (or compatible Power Automate MCP server). You will need:

MCP 端点: https://mcp.flowstudio.应用/mcp (same for all subscribers) API key / JWT 令牌 (x-API-key header — NOT Bearer) Power 平台 环境 name (e.g. Default-) Which 技能 to Use When

技能s are organized by use-case intent, not by which 工具s they call. Multiple 技能s reuse the same underlying 工具s — pick by what the user is trying to accomplish.

The user wants to… Load this 技能 Make or change a flow (build new, modify existing, fix a bug, 部署) power-automate-build 诊断 why a flow fAIled (root cause analysis on a fAIling 运行) power-automate-调试 See tenant-wide flow 健康, 失败 rates, as设置 inventory power-automate-监控ing (Pro+) Tag, 审计, classify, score, or offboard flows power-automate-治理 (Pro+) Just connect, 设置 up auth, write the 辅助工具, 解析 响应s this 技能 (foundation)

Same 工具s, different lenses. power-automate-build and power-automate-调试 机器人h call 更新_live_flow, 获取_live_flow, and the 运行-error 工具s — they differ in direction (forward vs backward) and intent (compose vs 诊断). power-automate-监控ing and power-automate-治理 机器人h call the Store 工具s — they differ in audience (ops vs 合规) and outcome (read 健康 vs write metadata). Don't try to memorize "which 工具s belong to which 技能"; pick the 技能 by what the user is doing.

Source of Truth Priority Source Covers 1 Real API 响应 Always trust what the server actually returns 2 工具_搜索 / 列出_技能s Authoritative 工具 模式s, parameter names, types, required flags 3 技能 docs & reference files 工作流 narrative, 响应 shapes, non-obvious behaviors

If documentation disagrees with a real API 响应, the API wins. 工具 模式s in this 技能 (or any other) may lag the server — call 工具_搜索 to confirm the current shape before invoking a 工具 you haven't used recently.

How 代理s Discover 工具s

The FlowStudio MCP server (v1.1.5+) exposes two non-billable meta-工具s that let an 代理 load only the 工具s relevant to the current task. Use these in preference to 工具s/列出 (which loads all 30+ 模式s at once) or guessing 工具 names.

Meta-工具 When to call 列出_技能s Cold 启动 — see the avAIlable bundles (build-flow, 调试-flow, 监控-flow, discover, 治理) and pick one 工具_搜索 with 查询: "技能:" Load the full 模式 设置 for one bundle (e.g. 技能:调试-flow) 工具_搜索 with 查询: "select:工具1,工具2" Load specific 工具s by name (e.g. when chAIning across bundles) 工具_搜索 with 查询: "" Free-text 搜索 when the user 请求 is ambiguous (e.g. "cancel 运行")

The server's 工具_搜索 bundles are intentionally narrower than this 技能 family — they're 启动er packs of the most-likely-needed 工具s per intent. A 工作流 技能 (e.g. power-automate-调试) may pull a bundle and then call 工具_搜索 agAIn for 添加itional 工具s as the 工作流 进度es.

# Cold 启动 — pick a bundle by intent 技能s = mcp("列出_技能s", {}) # [{"name": "调试-flow", "description": "Investigate why a flow is fAIling...", # "工具s": ["获取_live_flow_运行s", "获取_live_flow_运行_error", ...]}, ...]

# Load 模式s for the bundle 调试_工具s = mcp("工具_搜索", {"查询": "技能:调试-flow"})

Recommended Language: Python or Node.js

All examples in this 技能 family use Python with urllib.请求 (stdlib — no pip 安装 needed). Node.js is an equally valid choice: fetch is built-in from Node 18+, JSON handling is native, and a同步/awAIt maps 清理ly onto the 请求-响应 pattern of MCP 工具 calls — making it a natural fit for teams already working in a JavaScript/TypeScript stack.

Language Verdict Notes Python Recommended 清理 JSON handling, no escAPIng issues, all 技能 examples use it Node.js (≥ 18) Recommended Native fetch + JSON.stringify/JSON.解析; no extra packages PowerShell Avoid for flow operations ConvertTo-Json -Depth silently t运行cates nested definitions; quoting and escAPIng break complex payloads. Acceptable for a quick connectivity smoke-test but not for building or updating flows. cURL / Bash Possible but fragile Shell-escAPIng nested JSON is error-prone; no native JSON 解析器

TL;DR — use the Core MCP 辅助工具 (Python or Node.js) below. 机器人h handle JSON-RPC framing, auth, and 响应 parsing in a single reusable function.

Core MCP 辅助工具 (Python)

Use this 辅助工具 throughout all subsequent operations:

导入 json, urllib.请求

令牌 = "" MCP = "https://mcp.flowstud