How to Use Google Maps in Roo Code with AgentPatch
Building anything with a location component means you need real place data. Restaurant finders, store locators, business directories, travel apps. At some point you need actual business names, addresses, ratings, and hours. Hardcoding fake data wastes time. Sample datasets go stale. What you want is live data from Google Maps, pulled into your project without leaving the editor.
Roo Code is an open-source AI coding agent that runs as a VS Code extension. It supports MCP, which means it can call external tools during a conversation. Connect it to a Google Maps tool and it can look up businesses, addresses, and place details as part of whatever you’re building.
This is useful beyond location-specific apps too. Verifying a business address, looking up a timezone for a city, checking whether a storefront exists before building a feature around it. These small lookups come up regularly during development.
Setup
The AgentPatch CLI is designed for AI agents to use via shell access. Install it, and your agent can discover and invoke any tool on the marketplace.
Install (zero dependencies, Python 3.10+):
pip install agentpatch
Set your API key:
export AGENTPATCH_API_KEY=your_api_key
Example commands your agent will use:
ap search "web search"
ap run google-search --input '{"query": "test"}'
Get your API key from the AgentPatch dashboard.
Example: Seeding a Local Business Directory
You’re building a local business directory and need real data for development. You tell Roo Code:
“I need 5 real coffee shops in Seattle for the seed data. Grab them from Google Maps and put them in seed.ts.”
Roo Code searches Google Maps through AgentPatch and returns results like:
Searching Google Maps for "coffee shops Seattle"...
Found several results. Here are 5 for seed data:
1. Victrola Coffee Roasters, 310 E Pike St, Seattle, WA 98122 | Rating: 4.6
2. Espresso Vivace, 532 Broadway Ave E, Seattle, WA 98102 | Rating: 4.5
3. Caffe Vita, 1005 E Pike St, Seattle, WA 98122 | Rating: 4.4
4. Lighthouse Roasters, 400 N 43rd St, Seattle, WA 98103 | Rating: 4.7
5. Elm Coffee Roasters, 240 2nd Ave Ext S, Seattle, WA 98104 | Rating: 4.6
Writing these to seed.ts now.
Roo Code fetched live data and wrote the seed file. No tab-switching, no copy-pasting from a browser window.
What Roo Code Does Step by Step
When you ask for location data, Roo Code:
- Identifies that the request needs a Google Maps search.
- Calls the Maps tool through AgentPatch with your query and location parameters.
- Receives structured results: business names, addresses, ratings, review counts, categories.
- Uses the data however you asked, whether that’s writing it to a file, using it in a component, or just reporting it back.
Because Roo Code runs in VS Code, the results are immediately available in your workspace. It can write them to a TypeScript file, a JSON fixture, a database migration, or just display them in the chat panel.
Wrapping Up
Adding AgentPatch to Roo Code gives it live access to business and place data during any session. Google Maps is one tool. The same config unlocks web search, news, image generation, email, and every other tool on the marketplace. Visit agentpatch.ai to see what’s available.