{
  "markdown": "# FlowStudio MCP — Power Automate Skills for AI Agents\n\nGive your AI agent the same visibility you have in the Power Automate portal — plus a bit more.\nThe Graph API only returns top-level run status — agents can't see action inputs,\nloop iterations, or nested failures. FlowStudio MCP exposes all of it.\n\n![Agent debugging a Power Automate flow via MCP](assets/demo-debug.gif)\n\n**You can click through the portal and find the root cause. Your agent can't — unless it has MCP.**\n\n![The portal shows everything to a human — but agents only see the top-level error via Graph API](assets/portal-vs-reality.png)\n\n![With FlowStudio MCP, the agent sees what you see](assets/mcp-root-cause.png)\n\n## When you need this\n\n- Your agent can see that a flow failed, but not why — Graph API only returns status codes\n- You want your agent to see action-level inputs and outputs, like you can in the portal\n- A loop has hundreds of iterations and some produced bad output — in the portal you'd click through each one, but the agent can scan all iteration inputs and outputs at once\n- You want to check flow health, failure rates, and maker activity across your tenant without opening the admin center\n- You need to classify flows, detect orphaned resources, or audit connectors at scale — without installing the CoE Starter Kit\n- You're tired of being the middle-man between your agent and the portal\n\n## Graph API vs FlowStudio MCP\n\nThe core difference: **Graph API gives your agent run status. MCP gives your agent the inputs and outputs of every action.**\n\n| What the agent sees                           | Graph API | FlowStudio MCP |\n| --------------------------------------------- | --------- | --------------- |\n| Run passed or failed                          | Yes       | Yes             |\n| **Action inputs and outputs**                 | **No**    | **Yes**         |\n| Error details beyond status code              | No        | Yes             |\n| Child flow run details                        | No        | Yes             |\n| Loop iteration data                           | No        | Yes             |\n| Flow definition (read + write)                | Limited   | Full JSON       |\n| Resubmit / cancel runs                        | Limited   | Yes             |\n| **Cached flow health & failure rates**        | **No**    | **Yes**         |\n| **Maker / Power Apps / connection inventory** | **No**    | **Yes**         |\n| **Governance metadata (tags, impact, owner)** | **No**    | **Yes**         |\n\n## Skills\n\n| Skill                                                            | Description                                                                                                    |\n| ---------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------- |\n| [`power-automate-mcp`](skills/power-automate-mcp/)               | Connect to and operate Power Automate cloud flows — list flows, read definitions, check runs, resubmit, cancel |\n| [`power-automate-debug`](skills/power-automate-debug/)           | Step-by-step diagnostic process for investigating failing flows                                                |\n| [`power-automate-build`](skills/power-automate-build/)           | Build, scaffold, and deploy Power Automate flow definitions from scratch                                       |\n| [`power-automate-monitoring`](skills/power-automate-monitoring/) | Flow health, failure rates, maker inventory, Power Apps, environment and connection counts                     |\n| [`power-automate-governance`](skills/power-automate-governance/) | Classify flows by impact, detect orphans, audit connectors, manage notifications, compute archive scores       |\n\nThe first three skills use **live** Power Automate API calls. The monitoring and governance\nskills use the **cached store** — a daily snapshot with aggregated stats, remediation hints,\nand governance metadata. Requires a FlowStudio for Teams or MCP Pro+ subscription for store tools.\n\nEach skill follows the [Agent Skills specification](https://agentskills.io/specification)\nand works with any compatible agent.\n\n### Supported agents\n\nCopilot, Claude Code, ChatGPT, Codex, OpenClaw, Gemini CLI, Cursor, Goose, Amp, OpenHands\n\n## Guides\n\nFull walkthroughs on [learn.flowstudio.app](https://learn.flowstudio.app):\n\n- [Getting started](https://learn.flowstudio.app/mcp-getting-started) — install FlowStudio MCP for any agent\n- [Build flows with AI agents](https://learn.flowstudio.app/build-power-automate-flows-with-ai-agents) — a worked 3-step example\n- [Vibe code Power Automate](https://learn.flowstudio.app/vibe-code-power-automate) — describe the flow, the agent builds it\n- [Power Automate in VS Code](https://learn.flowstudio.app/power-automate-ide) — work on flows from your editor\n- [Debug a failing flow](https://learn.flowstudio.app/mcp-debug) — find the root cause from action-level data\n- By agent: [ChatGPT](https://learn.flowstudio.app/chatgpt-power-automate) · [Claude](https://learn.flowstudio.app/claude-power-automate) · [GitHub Copilot](https://learn.flowstudio.app/github-copilot-power-automate) · [Copilot Studio](https://learn.flowstudio.app/copilot-studio-power-automate) · [Codex](https://learn.flowstudio.app/codex-power-automate)\n\n## Quick Start\n\n### Connect from ChatGPT or claude.ai\n\nSign in with your Microsoft account instead of pasting a key. Add\n`https://mcp.flowstudio.app/mcp/oauth` as a connector in ChatGPT (Developer mode\non) or at claude.ai, then follow the sign-in prompt. Walkthrough:\n[ChatGPT for Power Automate](https://learn.flowstudio.app/chatgpt-power-automate).\n\n### Install as Claude Code plugin\n\nAvailable through the Claude plugin marketplace after approval. To test locally:\n\n```bash\ngit clone https://github.com/ninihen1/power-automate-mcp-skills.git\nclaude --plugin-dir ./power-automate-mcp-skills\n```\n\nThen connect the MCP server:\n\n```bash\nclaude mcp add --transport http flowstudio https://mcp.flowstudio.app/mcp \\\n  --header \"x-api-key: <YOUR_TOKEN>\"\n```\n\nGet your token at [mcp.flowstudio.app](https://mcp.flowstudio.app).\n\n### Install in Codex\n\nInside a Codex session, install skills directly:\n\n```\n$skill-installer install https://github.com/ninihen1/power-automate-mcp-skills/tree/main/skills/power-automate-mcp\n$skill-installer install https://github.com/ninihen1/power-automate-mcp-skills/tree/main/skills/power-automate-debug\n$skill-installer install https://github.com/ninihen1/power-automate-mcp-skills/tree/main/skills/power-automate-build\n$skill-installer install https://github.com/ninihen1/power-automate-mcp-skills/tree/main/skills/power-automate-monitoring\n$skill-installer install https://github.com/ninihen1/power-automate-mcp-skills/tree/main/skills/power-automate-governance\n```\n\nThen connect the MCP server in `~/.codex/config.toml`:\n\n```toml\n[mcp_servers.flowstudio]\nurl = \"https://mcp.flowstudio.app/mcp\"\n\n[mcp_servers.flowstudio.http_headers]\nx-api-key = \"<YOUR_TOKEN>\"\n```\n\n### Install via skills.sh\n\nSearch for [flowstudio on skills.sh](https://skills.sh/?q=flowstudio), or:\n\n```bash\nnpx skills add github/awesome-copilot -s flowstudio-power-automate-mcp\nnpx skills add github/awesome-copilot -s flowstudio-power-automate-debug\nnpx skills add github/awesome-copilot -s flowstudio-power-automate-build\nnpx skills add github/awesome-copilot -s flowstudio-power-automate-monitoring\nnpx skills add github/awesome-copilot -s flowstudio-power-automate-governance\n```\n\n### Install via ClawHub\n\n```bash\nnpx clawhub@latest install power-automate-mcp\n```\n\n### Install via Smithery\n\n```bash\nnpx smithery skill add flowstudio/power-automate-mcp\n```\n\n### Manual install\n\nCopy the skill folder(s) into your project's `.github/skills/` directory\n(or wherever your agent discovers skills).\n\n### Connect the MCP server\n\n**Claude Code:**\n\n```bash\nclaude mcp add --transport http flowstudio https://mcp.flowstudio.app/mcp \\\n  --header \"x-api-key: <YOUR_TOKEN>\"\n```\n\n**Codex** (`~/.codex/config.toml`):\n\n```toml\n[mcp_servers.flowstudio]\nurl = \"https://mcp.flowstudio.app/mcp\"\n\n[mcp_servers.flowstudio.http_headers]\nx-api-key = \"<YOUR_TOKEN>\"\n```\n\n**Copilot / VS Code** (`.vscode/mcp.json`):\n\n```json\n{\n  \"servers\": {\n    \"flowstudio\": {\n      \"type\": \"http\",\n      \"url\": \"https://mcp.flowstudio.app/mcp\",\n      \"headers\": { \"x-api-key\": \"<YOUR_TOKEN>\" }\n    }\n  }\n}\n```\n\nGet your token at [mcp.flowstudio.app](https://mcp.flowstudio.app).\n\n## Real debugging examples\n\nThese are from real production investigations, not demos.\n\n- **[Expression error in child flow](examples/fix-expression-error.md)** —\n  `contains(string(...))` crashed on a nested property. Agent traced through\n  parent flow, into child, through loop iterations, and found the failing input.\n  Portal showed \"ExpressionEvaluationFailed\" with no context.\n\n- **[Data entry, not a flow bug](examples/data-not-flow.md)** —\n  User reported two \"bugs\" back to back. Agent proved both were data entry\n  errors (missing comma in email, single address in CC field). Flow was correct.\n  Diagnosed in seconds.\n\n- **[Null value crashes child flow](examples/null-child-flow.md)** —\n  `split(Name, ', ')` crashed when 38% of records had null Names. Agent traced\n  parent to child to loop to action, found the root cause, and deployed a fix\n  via `update_live_flow`.\n\n## Prerequisites\n\n- A [FlowStudio](https://mcp.flowstudio.app) MCP subscription (all live tools)\n- For store tools (monitoring, governance): FlowStudio for Teams or MCP Pro+\n- MCP endpoint: `https://mcp.flowstudio.app/mcp`\n- API key / JWT token (passed as `x-api-key` header)\n\n## Repository structure\n\n```\nskills/\n  power-automate-mcp/          core connection & operation skill\n  power-automate-debug/        debug workflow skill\n  power-automate-build/        build & deploy skill\n  power-automate-monitoring/   flow health & tenant inventory skill\n  power-automate-governance/   compliance & governance skill\nexamples/                      real debugging walkthroughs\nREADME.md\nLICENSE                        MIT\n```\n\n## Available on GitHub\n\nWorks with Copilot, Claude, ChatGPT, and any MCP-compatible agent.\n\n- [awesome-copilot](https://github.com/github/awesome-copilot) (merged)\n- [skills.sh](https://skills.sh/?q=flowstudio) (3K+ installs)\n- [Smithery](https://smithery.ai/skills/flowstudio/power-automate-mcp) (published)\n- [ClawHub](https://clawhub.ai) (v1.1.0)\n\n## Contributing\n\nContributions welcome. Each skill folder must contain a `SKILL.md` with the\nrequired frontmatter. See the existing skills for the format.\n\n## License\n\n[MIT](LICENSE)\n\n---\n\nKeywords: Power Automate debugging, flow run history, expression evaluation failed,\nchild flow failure, nested action errors, loop iteration output, agent automation MCP,\nPower Platform AI, flow definition deploy, resubmit failed run, flow monitoring,\ngovernance, CoE, orphan detection, connector audit, archive score, maker inventory\n",
  "bytes": 10968,
  "sha": "6675ed035d298231cdfbd87a182e9f345505b7332c0ba7e2b8c1e735dbd99c3a",
  "repo_slug": "ninihen1/power-automate-mcp-skills",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_ninihen1_power_automate_mcp_skills_8304a9bd/readme"
}