TL;DR

Maps Intel is a free Claude skill that turns a one-line ask — “give me a competitive view of yoga studios in Lower Manhattan” — into a 36-column CSV plus a branded, interactive HTML dashboard. About five minutes end to end. macOS only. No API keys, no paid services.

• Built for: local SEO consultants, agencies, and in-house marketers doing competitor research on Google Maps
• Replaces: 1–2 days of manual tab-flipping, copy-pasting, and spreadsheet wrangling per market
• Requires: macOS, Docker Desktop (free), the Claude desktop app with Cowork mode
• Delivers: raw data plus a Will Do SEO-branded dashboard you can hand straight to a client

Why I built it

Every time I pick up a new local SEO client, the first week looks the same: pull every direct competitor off Google Maps, capture their review counts and ratings, re-run five to ten keyword variations to figure out who ranks where, identify the elite operators and the hidden gems, and then turn the whole thing into something the client or your team will actually look at.

That used to be a multi-hour slog per market. Tab after tab, copy-paste after copy-paste, then a separate design pass to make the spreadsheet pitch-worthy. The information lives on Google Maps. The pain point is pulling it at scale.

So I built a Claude skill that does it.

What Maps Intel actually does

Maps Intel is a Cowork skill — a packaged folder of instructions and helper scripts that Claude loads when the conversation calls for it. Once it’s installed, you describe the market in plain English. Claude:

1. Picks the keyword set (e.g. yoga-studio, hot-yoga, vinyasa-yoga, pilates-studio, barre-studio)
2. Confirms depth and whether you want full review text
3. Writes a self-contained, double-clickable .command file
4. You double-click. Docker spins up the gosom/google-maps-scraper container, runs the scrape on your Mac, then post-processes everything into a flat 36-column CSV and a branded interactive dashboard
5. Finder pops open to the run folder when it’s done. You open dashboard.html and you’re already at the deliverable

No Google Maps API key. No paid scraper service. No usage caps. The data lives on your machine.

Walkthrough: from idea to dashboard in five minutes

This is the fast tour. Real numbers from a real run.

Step 1 — Ask

“I need a competitive view of yoga, pilates, and barre studios in Lower Manhattan for a new client pitch.”

Claude reads the request, picks five keywords, asks me how deep per query (default ~60) and whether to pull review text (off by default — turning it on roughly triples the runtime). I confirm.

Step 2 — One file, one double-click

Claude generates yoga-fidi.command — a single self-contained shell script with the Docker invocation, the queries, and the post-process all baked in. It lands in my workspace folder. I double-click it.

Terminal opens. Docker pulls the gosom image (first run only, ~700MB, takes about five minutes on broadband). The scrape itself runs for another two to three minutes. The CSV flatten and dashboard build take another five seconds.

Step 3 — Open the dashboard

Finder pops open to the run folder automatically. Inside:

• results.json — raw NDJSON, one business per line
• results.csv — flat 36-column spreadsheet, every field preserved (nested data like reviews and hours stored as JSON strings in cells)
• dashboard.html — the deliverable

Double-click dashboard.html and you’re looking at a Will Do SEO-branded interactive report. Three tabs:

• Overview — KPI scorecards (total businesses, average rating, median reviews, elite operators), rating distribution chart, results-per-query bars, top 10 by review count, category mix doughnut
• Rankings Explorer — sortable, filterable table with rank-per-keyword baked in.
• Reviews Deep-dive — click any business, browse its sampled reviews on the right

What the data actually surfaces

For the FiDi yoga sweep above — 238 businesses across five keywords — the dashboard answered the questions I’d usually spend the second day of research on:

• YogaRenew is the giant of the market — 1,401 reviews, 5.0★. Hard to compete with on volume.
• Sui Yoga & Spa Thermae owns the hot-yoga query at #1 with 1,114 reviews. If a client is opening a hot-yoga studio in FiDi, this is who they’re up against.
• VERAYOGA — 5.0★, only 48 reviews, ranked #55 for hot-yoga. That’s a hidden gem. Probably newer, probably better than its review count suggests, probably worth watching.
• 45 elite operators (4.8★ with 100+ reviews) — 19% of the market. That’s the competitive set.

Same flow for a different market: I ran “dentists in and around South Kingstown, RI” as a test, six geo queries, ~8 minutes end to end, 96 unique dentists across the area.

Who this is for

• Local SEO consultants doing weekly or monthly competitor reports for clients
• Agencies running new-client onboarding research at scale
• In-house marketers at multi-location businesses who need to understand each market before launching
• Anyone who’s typed “every [business type] in [zip code]” into a spreadsheet by hand more than once

What you need

• macOS — the .command file flow is Mac-specific
• Docker Desktop for Mac — free from docker.com. Required because the actual scrape runs in a Docker container
• The Claude desktop app with Cowork mode

That’s the whole list.

Install

First — install Docker Desktop (skip if you already have it)

The scrape itself runs in a Docker container, so Docker Desktop has to be installed and running before the skill can do anything. It’s free.

1. Download Docker Desktop for Mac from docker.com. The installer is ~700MB and takes a couple of minutes.
2. Open Docker Desktop after install. Wait for the whale icon in your menu bar to stop animating — that means the daemon is up.
3. Leave Docker Desktop running. The skill will use it automatically whenever you ask Claude for a scrape.

First-time setup including the Docker download takes about 5–10 minutes total. After that it’s a one-time thing — Docker just lives on your Mac.

Then — install the Maps Intel skill (about 30 seconds)

1. Download the .skill file from the link at the bottom of this post
2. Open Claude, switch to Cowork mode
3. Go to customize → create new skill, click the plus, then upload skill
4. Click Save skill. Done.

The skill installs itself into Claude’s skills folder. Next time you describe a local market in a conversation — even casually — Claude picks it up automatically.

Under the hood

Maps Intel is a zip file with a few pieces:

• SKILL.md — the instructions Claude follows
• scripts/ — Python helpers: the .command builder, the dashboard generator, the web-UI command builder
• references/ — every gosom CLI flag and the full output schema, loaded by Claude on demand
• assets/ — the Will Do SEO logo (PNG plus a pre-computed base64 data URI for inline embedding)

The dashboard’s branding is baked into the skill — palette, logo, tagline — so anyone who installs it gets Will Do SEO branded output without any setup. If you want to fork it for your own brand, unzip the .skill file, swap the logo and palette in scripts/build_dashboard.py, zip it back up, and re-install. Fully open.

Credit where it’s due

The actual scraping engine is gosom/google-maps-scraper — a brilliant open source Go project by Giorgos Komninos. MIT-licensed, free, and the engineering behind it is genuinely impressive. If you find Maps Intel useful, go star the gosom repo — that’s where the foundation lives.

FAQ