The most lacking capability for AI Agents in the Chinese market right now is the ability to access real local data. AMap (Gaode Maps) has just opened a gateway: amap-lbs-skill, adapted for the OpenClaw platform, MIT license, with 9 stars. It packs five capabilities:

• POI search
• Route planning (walking/driving/cycling/public transit)
• Intelligent travel planner
• Nearby search
• Heatmap data visualization

The invocation follows the Skill standard format: SKILL.md, scripts/, index.js — just configure an AMap Web Service Key and it runs. Any Agent that uses the Claude Skill or OpenClaw Skill interface can connect directly. Another major Chinese tech company has stepped into providing Skill integrations.

Those five capabilities, when connected to an Agent, enable real-world scenarios:

• POI search → “List all Michelin restaurants in Shanghai”, “Top 10 cafes near 798”, “How many pediatric hospitals within 1km of my home?”
• Route planning → “Visit 5 clients in Beijing tomorrow, optimize route by shortest time”, “Best route for a one-day West Lake tour without getting tired”, “From Pudong airport to The Bund, driving vs subway — which is faster?”
• Intelligent travel planner → “3-day trip in Chengdu, give me attractions + food + route + map links”, “48-hour Tokyo weekend, budget ¥5000, create itinerary”
• Nearby search → “Any convenience stores within 500m of the hotel?”, “Nearest EV charging station to my current location”, “A café near the meeting room that allows drive-in”
• Heatmap visualization → “Generate a heatmap of pet hospitals in Shanghai”, “Plot customer data on map to see which area has highest sales density”, “Highlight top 10 neighborhoods by convenience store density”

Combined, these five capabilities mean one thing: AI Agents in mainland China can finally “get things done” and produce visual results. Travel guides upgrade from plain text to visual map links with POIs and routes, ready to share.

http://github.com/AMap-Web/amap-lbs-skill…

AMap-Web/amap-lbs-skill

Source: https://github.com/AMap-Web/amap-lbs-skill

AMap LBS Skill

AMap LBS Skill provides developers with comprehensive map data services, including location search, route planning, travel planning, and data visualization.

Features

• ✅ Automatic AMap Web Service Key management
• ✅ POI search
• ✅ Route planning (walking, driving, cycling, public transit)
• ✅ Intelligent travel planner
• ✅ Map visualization link generation
• ✅ Heatmap data visualization
• ✅ Command-line script execution
• ✅ Persistent local configuration

Install dependencies

npm install

Configure API Key

You need to configure an AMap Web Service Key on first use:

# Method 1: via environment variable at runtime
export AMAP_WEBSERVICE_KEY=your_key
node scripts/poi-search.js --keywords=KFC --city=Beijing

# Method 2: runtime prompt to input (saved to config.json)
node scripts/poi-search.js --keywords=KFC --city=Beijing

# Method 3: manually create configuration file
cp config.example.json config.json
# then edit config.json and fill in your Key

Get API Key: Visit the AMap Open Platform (https://lbs.amap.com/api/webservice/create-project-and-key) to create an application and obtain a Key.

Usage

1. POI Search

# Basic search
node scripts/poi-search.js --keywords=KFC --city=Beijing

# Search with more parameters
node scripts/poi-search.js --keywords=Restaurant --city=Shanghai --page=1 --offset=20

# Nearby search (requires center point coordinates and radius)
node scripts/poi-search.js --keywords=Hotel --location=116.397428,39.90923 --radius=1000

2. Route Planning

# Walking route
node scripts/route-planning.js --type=walking --origin=116.397428,39.90923 --destination=116.427281,39.903719

# Driving route (with waypoints)
node scripts/route-planning.js --type=driving --origin=116.397428,39.90923 --destination=116.427281,39.903719 --waypoints=116.410000,39.910000

# Cycling route
node scripts/route-planning.js --type=riding --origin=116.397428,39.90923 --destination=116.427281,39.903719

# Public transit route
node scripts/route-planning.js --type=transfer --origin=116.397428,39.90923 --destination=116.427281,39.903719 --city=Beijing

3. Intelligent Travel Planner

# Basic travel planning
node scripts/travel-planner.js --city=Beijing --interests=Attractions,Food,Hotels

# Specify route type
node scripts/travel-planner.js --city=Hangzhou --interests=West Lake,Food,Tea House --routeType=walking

# Driving tour
node scripts/travel-planner.js --city=Shanghai --interests=The Bund,Nanjing Road,Chenghuang Temple --routeType=driving

4. Use in Code

const { searchPOI, walkingRoute, drivingRoute, travelPlanner, generateMapLink } = require('./index');

// POI Search
async function searchExample() {
    const result = await searchPOI({ keywords: 'KFC', city: 'Beijing', page: 1, offset: 10 });
    console.log(result);
}

// Walking route planning
async function routeExample() {
    const result = await walkingRoute({ origin: '116.397428,39.90923', destination: '116.427281,39.903719' });
    console.log(result);
}

// Travel planning
async function travelExample() {
    const result = await travelPlanner({ city: 'Beijing', interests: ['Attractions', 'Food', 'Hotels'], routeType: 'walking' });
    console.log(result.mapLink); // Map visualization link
}

// Generate map link
function mapLinkExample() {
    const mapData = [
        { type: 'poi', lnglat: [116.397428, 39.90923], sort: 'Scenic Spot', text: 'Forbidden City', remark: 'Imperial palace of Ming and Qing dynasties' },
        { type: 'route', routeType: 'walking', start: [116.397428, 39.90923], end: [116.427281, 39.903719], remark: 'Walking route' }
    ];
    const link = generateMapLink(mapData);
    console.log(link);
}

API Parameter Description

POI Search Parameters

Route Planning Parameters

Walking Route (walkingRoute)

Driving Route (drivingRoute)

Cycling Route (ridingRoute)

Public Transit Route (transitRoute)

Travel Planner Parameters (travelPlanner)

Project Structure

jsapi-skills/
├── index.js                 # Main entry file, core functions
├── scripts/
│   ├── poi-search.js        # POI search script
│   ├── route-planning.js    # Route planning script
│   └── travel-planner.js    # Intelligent travel planner script
├── config.json              # Configuration file (auto-generated, do not commit)
├── config.example.json      # Configuration example
├── package.json             # Dependency configuration
├── .gitignore               # Git ignore configuration
├── SKILL.md                 # OpenClaw Skill description file
└── README.md                # This file

Map Visualization

All planning results generate a map visualization link in the following format:

https://a.amap.com/jsapi_demo_show/static/openclaw/travel_plan.html?data=

The data format follows the MapTaskData interface specification, supporting:

• POI tasks: display point of interest locations and information
• Route tasks: display route planning results

Example data structure:

[
    // POI point of interest
    {
        type: 'poi',
        lnglat: [116.397428, 39.90923],
        sort: 'Scenic Spot',
        text: 'Forbidden City',
        remark: 'Imperial palace of Ming and Qing dynasties, formerly known as the Forbidden City.'
    },
    // Route planning
    {
        type: 'route',
        routeType: 'walking',
        start: [116.397428, 39.90923],
        end: [116.427281, 39.903719],
        remark: 'Walking route'
    }
]

Notes

1. Keep your Web Service Key secure; do not commit it to public repositories.
2. config.json is already in .gitignore and will not be committed.
3. AMap Web Service API has call frequency limits; use reasonably.
4. Free users have daily call limits; see the AMap Open Platform documentation for details.

Related Links

• AMap Open Platform (https://lbs.amap.com/)
• Create application and get Key (https://lbs.amap.com/api/webservice/create-project-and-key)
• POI Search API documentation (https://lbs.amap.com/api/webservice/guide/api-advanced/newpoisearch)
• Web Service API overview (https://lbs.amap.com/api/webservice/summary)

License

MIT

Jason Zhu (@GoSailGlobal):
That’s quite something

Alibaba’s AMap (Gaode Maps), someone actually submitted amap-lbs-skill to my little site

Two of the three submitted projects were accepted:

☑️ https://t.co/0V34mJKxgc
☑️ https://t.co/XluGaMH5Ay
❌