StackMCP is the execution layer for MCP workflows.

Run, trace, audit, and reuse tool workflows across CLI, agents, local apps, and MCP servers.

Install

npm install -g stackmcp

YouTube

action fetch transcript

AI Summary

prompt summarize this transcript

Obsidian Note

target Everything/Web/

youtube-to-obsidian Name

3 Steps

~1 min Est. duration

Settings

Stop on Error Stop execution if any step fails

Retry Failed Steps Retry failed steps up to 3 times

Timeout Set a timeout for each step

3 mins

stackmcp run youtube-to-obsidian

A premade stack in the Stack Builder — drag, connect, configure tools. Install and try it in seconds.

How it works

Define

Compose tools and APIs into a workflow stack. Each stack is a reusable JSON template.

Run

Trigger stacks from your terminal or AI assistant. The daemon handles all MCP server connections.

Export any stack as a shareable template. Use community stacks or publish your own.

Why StackMCP

Local-first

Runs on your machine. No cloud proxy, no data leaving your environment unless you route it.

Uses MCP

Built on the Model Context Protocol. Works with any MCP-compatible AI assistant.

Reusable

Stacks are portable JSON. Share them, version them, build on existing ones.

Get started in seconds

Install globally, configure your MCP servers, and run your first stack.

Terminal

$ stackmcp init
? Enable the Obsidian MCP server? Yes
? Enable the YouTube MCP server? Yes
✓ Servers configured

$ stackmcp stacks
  youtube-to-obsidian  Fetch transcript, summarize, save to vault
  ...

$ stackmcp run youtube-to-obsidian \
    --url "https://youtube.com/watch?v=..."
✓ Stack complete — note saved to Obsidian