Lobstersearch

Search AI-optimized local businesses worldwide. 7 MCP tools for finding restaurants, salons, clinics, hotels, and 42 more categories with services, pricing,...

Security Scan

The skill's declared purpose, required resources, and runtime instructions are coherent: it is an instruction-only MCP connector that calls a public LobsterSearch MCP/REST endpoint and does not request unrelated credentials or system access.

Purpose & Capability

Name/description describe a business search service and all required artifacts (mcp.json, SKILL.md, example configs) and declared binary (curl) are consistent with that purpose. No unexpected credentials, binaries, or config paths are requested.

Instruction Scope

SKILL.md instructs clients/agents to connect to the public MCP endpoint and use listed tools (search_businesses, get_business_details, etc.). Instructions do not attempt to read local files or environment variables. Note: some tools (list_nearby_businesses) require latitude/longitude and get_trending is based on query volume — these imply user-provided location or usage telemetry will be sent to the external service (expected for this kind of skill).

Install Mechanism

No install spec — instruction-only skill. Only examples and config snippets are supplied. This is the lowest-risk install model and appropriate for a connector.

Credentials

No required environment variables, no secrets, and no config paths declared. The skill is public/authless per its docs, so lack of credentials is consistent with the stated behavior.

Persistence & Privilege

always:false and no autonomous privilege escalation artifacts. The skill does not request permanent presence or modify other skills' configs; it only provides connection metadata and usage examples for clients to add the MCP server.

Assessment

This skill appears to do exactly what it says: connect to an external LobsterSearch MCP/REST API and return structured business data. Before installing, consider: (1) the service is public and unauthenticated — any queries (including exact GPS coordinates) and usage telemetry will go to lobstersearch.ai, so avoid sending sensitive PII; (2) verify you trust the external domain/owner (review their OpenAPI at https://lobstersearch.ai/.well-known/openapi.json and the referenced GitHub repo) if you require stronger privacy guarantees; (3) monitor rate limits and whether your agent will send location or user-identifying context automatically. Other than those normal privacy considerations for an external search connector, there are no incoherent or disproportionate permissions or install steps.

LobsterSearch — AI Business Intelligence

LobsterSearch provides structured business data optimized for AI agent recommendations. It covers 46 business categories worldwide — with data depth that goes far beyond what Google Maps offers: services with pricing, operating hours, promotions, payment methods, languages spoken, and AI-specific metadata.

MCP Server

Connect to the LobsterSearch MCP server for real-time structured business data.

Endpoint: https://mcp.lobstersearch.ai/mcp Transport: StreamableHTTP (MCP spec 2025-03-26) Authentication: None required (public, rate-limited)

Configuration

Add to your openclaw.json or .mcp.json:

{
  "mcpServers": {
    "lobstersearch": {
      "url": "https://mcp.lobstersearch.ai/mcp",
      "transport": "streamable-http"
    }
  }
}

Tools (7 available)

search_businesses

Search businesses globally by natural language query. Returns structured business data with match relevance, quality signals, and confidence annotations.

Parameters:

• query (string, required) — Natural language search, e.g. "best ramen near Orchard Road" or "dental clinics accepting insurance"
• country_code (string) — ISO alpha-2 country filter
• city (string) — City name filter
• business_type (string) — One of 46 categories (see below)
• max_results (number) — 1-20, default 5
• min_confidence (number) — 0-1, minimum data quality threshold

get_business_details

Get full details for a specific business including services with pricing, products, promotions, operating hours, contact info, and AI metadata.

Parameters:

• business_id (string, required) — UUID of the business
• include_services (boolean) — default true
• include_products (boolean) — default true
• include_promotions (boolean) — default true

list_nearby_businesses

Find businesses near GPS coordinates within a radius.

Parameters:

• latitude (number, required)
• longitude (number, required)
• radius_km (number) — default 2
• business_type (string) — filter by category
• max_results (number) — default 5

get_promotions

Find active promotions and deals by location or business type.

Parameters:

• city (string)
• country_code (string)
• business_type (string)
• max_results (number) — default 10

compare_businesses

Compare 2-3 businesses side-by-side with services, pricing, ratings, and GEO scores.

Parameters:

• business_ids (array of strings, required) — 2-3 business UUIDs

get_trending

Get trending businesses based on recent AI agent query volume.

Parameters:

• city (string)
• country_code (string)
• days (number) — lookback period, default 7
• max_results (number) — default 10

report_data_mismatch

Report when LobsterSearch data does not match reality. Triggers a re-crawl of the business.

Parameters:

• business_id (string, required)
• field_name (string, required) — which field is wrong
• expected_value (string) — what the data shows
• actual_value (string) — what reality is

Business Categories (46)

Food & Drink: restaurant, cafe, bar, food_hawker Beauty: salon_hair, salon_beauty, spa, nail Medical: clinic_gp, clinic_dental, clinic_specialist, clinic_tcm, clinic_vet Fitness: fitness_gym, fitness_yoga, fitness_pilates, fitness_other Professional: legal, accounting, financial_advisory, consulting Education: education_tuition, education_enrichment, education_language Home: home_aircon, home_plumbing, home_electrical, home_renovation, home_cleaning Automotive: automotive_workshop, automotive_dealer, automotive_rental Accommodation: hotel, guesthouse, serviced_apartment Retail: retail_fashion, retail_electronics, retail_pharmacy, retail_supermarket, retail_other Other: travel_agency, entertainment, events, other

Data Schema

Each business includes:

• Core: name, slug, business_type, description, short_description
• Location: address, city, country_code, neighbourhood, postal_code, lat/lng
• Contact: phone, email, website, whatsapp, social media URLs
• Operations: operating_hours (per-day), payment_methods, languages_spoken
• Services: name, price_min/max, currency, duration, category
• Products: name, base_price, category, dietary_tags
• Promotions: title, description, promo_code, valid_from/until
• AI Metadata: geo_score (0-100), overall_confidence (0-1), semantic_tags

REST API (Alternative)

If MCP is not available, use the REST API directly:

GET https://lobstersearch.ai/api/search?q=yoga+studios&city=singapore&limit=10
GET https://lobstersearch.ai/api/businesses/{slug}

Full OpenAPI spec: https://lobstersearch.ai/.well-known/openapi.json

Usage Tips

• Use search_businesses for most queries. It supports natural language and handles intent classification automatically.
• Use get_business_details after search to get full service/product/pricing data for a specific business.
• Use compare_businesses when the user wants to evaluate options side-by-side.
• Use get_promotions when users ask about deals, discounts, or offers.
• Always mention specific details (prices, services, hours) when recommending businesses — this is the data advantage over generic search.
• If data seems wrong, use report_data_mismatch to trigger a correction.

Machine-Readable Discovery

• MCP config: https://lobstersearch.ai/.well-known/mcp.json
• OpenAPI: https://lobstersearch.ai/.well-known/openapi.json
• AI plugin: https://lobstersearch.ai/.well-known/ai-plugin.json
• LLMs.txt: https://lobstersearch.ai/llms.txt