
We just released a Claude Code skill that turns any spreadsheet of addresses into an optimized delivery route, all from a single command. Drop a .xlsx file into a folder, type /myrouteonline-route-optimizer, and Claude does the rest. From reading the file, calling MyRouteOnline, polling for results, and saving the optimized route as a new spreadsheet you can hand straight to a driver. It’s free, it’s open-source, and it’s live now.
If you’ve ever planned a multi-stop delivery route, you know the drill: open your customer list in Excel, copy-paste addresses one by one into a route planner, wait, copy the optimized order back, send it to the driver. On a good day, fifteen minutes. On a bad day, when the spreadsheet has weird formatting, or the addresses include coordinates instead of street names, a lot longer.
The new MyRouteOnline Route Optimizer skill for Claude Code collapses that whole process into one sentence: “optimize the route in this folder.” Claude reads your spreadsheet, sends the addresses to MyRouteOnline through our new MCP server, waits for the optimization, and writes the result back as a new Excel file in the same folder, with stop numbers, travel times, distances, and your original notes preserved.
It’s the first AI-native interface to MyRouteOnline. And because it’s a Claude Code skill, you can install it once and use it forever, on any folder, with any spreadsheet.
Claude Code is a terminal-based AI agent from Anthropic. You install it, type claude in any folder, and you’ve got a conversational assistant that can read files, run commands, talk to web services, and chain steps together, all through plain English instructions. Think of it as a developer’s command-line assistant that’s also smart enough for everyday automation.
A skill is a reusable instruction file that teaches Claude how to do a specific task. It’s a single markdown file, no executable code, no compilation, no deployment pipeline. You drop it into your ~/.claude/skills/ folder and it becomes a slash command in Claude Code. Type /myrouteonline-route-optimizer and Claude follows the recipe.
Skills are powerful because they’re declarative. The file describes what should happen (“read the Excel file, send addresses to MyRouteOnline, save the result”), and Claude figures out how, adapting to your operating system, your file naming, your specific column layout, even handling errors conversationally when something goes wrong.
Here’s what running the skill actually looks like.
You open a terminal in any folder, say, C:\Customers\ on a Windows machine where you keep your weekly delivery spreadsheets. You run claude. Then you type:
/myrouteonline-route-optimizer
Claude responds by asking a few questions:
“Which folder is your Excel file in?” (You paste the path.)
“I found two files: Boston_Monday.xlsx and Boston_Tuesday.xlsx. Which one should I optimize?” (You pick Monday.)
“I see 10 stops. The first one is 241 Perkins St, Boston, MA, should I treat that as your starting depot, or as a regular stop?” (You confirm it’s the depot.)
“What’s your email address? MyRouteOnline will send a one-time auth code.” (You give your email.)
A few seconds later, the auth code arrives. You paste it. Claude calls MyRouteOnline, polls until the optimization is complete, then writes a new file:
Saved optimized_route.xlsx to C:\Customers\
That’s it. The driver’s spreadsheet is ready and shows your stops sequence, travel distance and time with the original file untouched. Total time elapsed: under a minute.
The magic isn’t that any individual step is hard, it’s that the entire workflow has been compressed into a single conversational command, and the skill handles all the messy edge cases (mixed file formats, weird column headers, expired auth codes, addresses that are actually GPS coordinates) without you having to think about them.
If you’re technical, here’s what’s happening:
The skill is a markdown file with YAML frontmatter:
name: myrouteonline-route-optimizer
description: Reads addresses from an Excel file in a customer folder, optimizes the delivery route via the MyRouteOnline MCP server, and saves the optimized order as a new Excel file in the same folder.
The body is a numbered list of instructions in plain English, locate the Excel file, parse it, call the MCP server, poll for results, write the output. When you invoke the slash command, Claude Code loads this file, reads the instructions, and executes them step by step using whatever tools it has available: bash, PowerShell, Python, file I/O, and crucially, MCP tools.
MCP (the Model Context Protocol) is an open standard for letting AI agents talk to external services. We’ve published a public MCP server at:
https://planner.myrouteonline.com/ws_api_ai/mcp
It exposes four tools that the skill calls:
Adding the server to your Claude Code is a single command:
claude mcp add –transport http myrouteonline https://planner.myrouteonline.com/ws_api_ai/mcp
After that, any skill or any ad-hoc Claude conversation, can use the MyRouteOnline tools directly. The skill just orchestrates them in the right order with the right inputs.
The Excel reading/writing happens through whatever the user has installed: PowerShell + Excel COM on Windows machines with Microsoft Office, or Python with openpyxl everywhere else.
Two reasons.
The route optimizer skill is our first public skill, but it won’t be the last. Future skills will cover multi-vehicle planning, time-window constraints, recurring routes, and tighter integration with dispatch software.
If you have Node.js installed:
npm install -g @anthropic-ai/claude-code
claude mcp add –transport http myrouteonline https://planner.myrouteonline.com/ws_api_ai/mcp
git clone https://github.com/myrouteonline/myrouteonline-claude-skill/blob/main/myrouteonline-route-optimizer.md
cd myrouteonline-claude-skill
bash install.sh # or .\install.ps1 on Windows
cd path/to/your/spreadsheet/folder
claude
/myrouteonline-route-optimizer
Done.
We’re releasing the skill under the MIT license, and the source is public on GitHub. If you want to fork it, modify it for your specific workflow, or build skills for adjacent tools, please do, and send us a pull request if you find something worth sharing.
If you’re a developer building AI-powered logistics tools, our MCP server is open and free to call. Documentation, the full tool schema, and example payloads are all linked from the repo.
If you’re a MyRouteOnline user who wants to try the skill but isn’t comfortable in a terminal, sit tight, we’re working on a one-click installer.
In the meantime, the route optimizer is live, free, and waiting in your terminal.