Developers / Agent Skill

Give your agent a better starting point.

A reusable playbook for Google Maps data. Teach your agent how to run one search, read its JSON response, and turn business listings into useful results.

01 / Context your agent can use

The connection is only the beginning.

MCP supplies tools. The skill explains how to use them together, with a REST fallback when a configured MCP connection is unavailable.

01

Know which tool to use

Operation references and result schemas help the agent select the right request and interpret the response.

02

Finish the workflow

Instructions cover 20-business responses, source pages, safe retries, and missing data.

03

Keep access under control

Credentials stay in your runtime’s secret configuration. The skill carries no embedded key or subscription.

Inside the package

  • SKILL.md — workflow instructions
  • Operation references and normalized result schemas
  • Agent metadata for compatible runtimes
  • Dependency-free REST request script
  • Installer with checksum verification
  • Manifest listing the complete package

02 / Install once

A small package. A repeatable workflow.

Requires Node.js 24 and an agent runtime that can load skills from a local directory. No API key is needed to download or install the package.

  1. 01

    Download and review

    Save the installer locally and read it before running it. Review the skill instructions and manifest as well.

  2. 02

    Choose a new directory

    Use your runtime’s supported skill location. The installer downloads the complete package, verifies checksums, and refuses to overwrite an existing installation.

  3. 03

    Connect your credentials

    Prefer your configured MCP connection. For REST fallback, inject GMSCRAPER_API_KEY through the runtime environment.

Terminal · installation

Replace /path/to/skills with your runtime’s skill directory. Review the downloaded script before the node command.

curl --fail --silent --show-error \
  https://gmapscrawl.com/skills/google-maps-data/scripts/install.mjs \
  --output /tmp/install-google-maps-data.mjs

# Review the installer, then install into a NEW directory:
node /tmp/install-google-maps-data.mjs \
  /path/to/skills/google-maps-data

03 / Try it out

Start small. Follow the result.

Enable the installed skill in your agent runtime and validate setup with a test credential. Test data is simulated; live searches require API entitlement and service availability.

Your first workflow
Use the Google Maps Data skill and my configured test credential.
Search one page for coffee shops in Seattle.
Read the businesses returned in the response.
Show names, websites, and ratings. Label all test data as simulated.
Do not expand the search or make additional requests.

Frequently asked questions

Does installing the skill give me API access?

No. The package contains instructions, references, and helper scripts. API and MCP calls still require your workspace credentials, scopes, and entitlement.

Is downloading SKILL.md enough?

No. The skill relies on its scripts and references. Use the installer to retrieve and verify the complete package.

How do I update an existing installation?

Install into a new directory, review the changes, then replace the old installation using your runtime’s skill management workflow. The installer does not overwrite an existing directory.

Can the skill retrieve individual reviews or photos?

Standalone review and photo jobs are not currently supported. The agent should report unavailable capabilities and stay within the requested business-search workflow.

Take the next step

Connect the tools behind the skill.

Set up the remote MCP endpoint, configure your API credential, and let your agent discover the available tool schemas.