Mapbox Search Integration logo

Mapbox Search Integration

Organization
mapbox
mapbox-search-integration

Complete workflow for implementing Mapbox search in applications - from discovery questions to production-ready integration with best practices

Overview

Publishermapbox
Repositorymapbox-agent-skills
Skill namemapbox-search-integration
Stars
78
Forks
17
Bundled files
10
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 10 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by mapbox on GitHub. Read the source before you install it.

Installation

Install the Mapbox Search Integration AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/mapbox/mapbox-agent-skills.git /tmp/mapbox-agent-skills
mkdir -p .claude/skills
cp -r /tmp/mapbox-agent-skills/skills/mapbox-search-integration .claude/skills/mapbox-search-integration
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Mapbox Search Integration in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Mapbox Search Integration on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Mapbox Search Integration is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Mapbox Search Integration Skill

Expert guidance for implementing Mapbox search functionality in applications. Covers the complete workflow from asking the right discovery questions, selecting the appropriate search product, to implementing production-ready integrations following best practices from the Mapbox search team.

Use This Skill When

User says things like:

  • "I need to add search to my map"
  • "I need a search bar for my mapping app"
  • "How do I implement location search?"
  • "I want users to search for places/addresses"
  • "I need geocoding in my application"

This skill complements mapbox-search-patterns:

  • mapbox-search-patterns = Tool and parameter selection
  • mapbox-search-integration = Complete implementation workflow

Discovery Phase: Ask the Right Questions

Before jumping into code, ask these questions to understand requirements:

Question 1: What are users searching for?

Ask: "What do you want users to search for?"

Common answers and implications:

  • "Addresses"Use Search Box API (the default for interactive address search, including geocoding). Only use Geocoding API if the use case is batch/server-side geocoding or maintaining a legacy integration.
  • "Points of interest / businesses" → POI search, use Search Box API with category search
  • "Both addresses and POIs" → Search Box API
  • "Specific types of POIs" (restaurants, hotels, etc.) → Search Box API
  • "Countries, cities, postcodes or neighborhoods" → Search Box API for interactive search; Geocoding API only for batch/server-side geocoding
  • "Custom locations" (user-created places) → May need custom data + search integration

Follow-up if not stated initially: "Are your users searching for points of interest data? Restaurants, stores, categories of businesses?"

Implications:

  • "Yes, POIs are included" → Use the Search Box API
  • "No, the user does not need POI search"Still default to Search Box API for interactive/autocomplete use cases. Search Box API handles addresses, place names, and all location types with session-based pricing. Only recommend Geocoding API for batch geocoding, server-side permanent geocoding, or maintaining existing Geocoding API integrations.

Question 2: What's the geographic scope?

Ask: "Where will users be searching?"

Common answers and implications:

  • "Single country" (e.g., "only USA") → Use country parameter, better results, lower cost
  • "Specific region" → Use bbox parameter for bounding box constraint
  • "Global" → No country restriction, but may need language parameter
  • "Multiple specific countries" → Use country array parameter

Follow-up: "Do you need to limit results to a specific area?" (delivery zone, service area, etc.)

Question 3: What's the search interaction pattern?

Ask: "How will users interact with search?"

Common answers and implications:

  • "Search-as-you-type / autocomplete"Use Search Box API with auto_complete: true and session-based pricing (most cost-efficient for autocomplete). Implement debouncing.
  • "Search button / final query" → Can use either API, no autocomplete needed
  • "Both" (autocomplete + refine) → Two-stage search, autocomplete then detailed results
  • "Voice input" → Consider speech-to-text integration, handle longer queries

Question 4: What platform?

Ask: "What platform is this for?"

Common answers and implications:

  • "Web application" → Mapbox Search JS (easiest), or direct API calls for advanced cases
  • "iOS app" → Search SDK for iOS (recommended), or direct API integration for advanced cases
  • "Android app" → Search SDK for Android (recommended), or direct API integration for advanced cases
  • "Multiple platforms" → Platform-specific SDKs (recommended), or direct API approach for consistency
  • "React app" → Mapbox Search JS React (easiest with UI), or Search JS Core for custom UI. Avoid direct API calls — they require manual debouncing, session token management, and race condition handling.
  • "Vue / Angular / Other framework" → Mapbox Search JS Core or Web. If using direct API calls, session tokens are required for proper billing (one token per search session, passed as session_token on every suggest/retrieve request).

Question 5: How will results be used?

Ask: "What happens when a user selects a result?"

Common answers and implications:

  • "Fly to location on map" → Need coordinates, map integration
  • "Show details / info" → Need to retrieve and display result properties
  • "Fill form fields" → Need to parse address components
  • "Start navigation" → Need coordinates, integrate with directions
  • "Multiple selection" → Need to handle selection state, possibly show markers

Question 6: Expected usage volume?

Ask: "How many searches do you expect per month?"

Implications:

  • Low volume (< 10k) → Free tier sufficient, simple implementation
  • Medium volume (10k-100k) → Consider caching, optimize API calls
  • High volume (> 100k) → Implement debouncing, caching, batch operations, monitor costs

Product Selection Decision Tree

Based on discovery answers, recommend the right product:

Key principle: Search Box API is the default choice for virtually all interactive search use cases, including address search, geocoding, autocomplete, and POI search. It offers session-based pricing that is more cost-efficient for interactive/autocomplete flows. Only recommend Geocoding API for the narrow cases listed below.

Search Box API (DEFAULT)

Use when (any of these):

  • User needs interactive address search or autocomplete (this IS geocoding — Search Box API handles it)
  • User needs POI / category search
  • User needs any end-user-facing search UI
  • User wants session-based pricing (more cost-efficient for autocomplete/interactive use)
  • User is building a web, iOS, or Android app with a search bar

Prefer SDKs over direct API calls for web integration:

  • Mapbox Search JS (SDK) - Recommended for web integration, with three components:
    • Search JS React - Easy search integration via React library with UI
    • Search JS Web - Easy search integration via Web Components with UI
    • Search JS Core - JavaScript (node or web) wrapper for API, build your own UI
  • Search Box API (REST) - Direct API integration, for advanced/custom cases
  • Search SDK for iOS - Native iOS integration
  • Search SDK for Android - Native Android integration

Geocoding API (SPECIALIZED)

Use ONLY when:

  • Batch geocoding large lists of addresses (server-side)
  • Permanent/stored geocoding results (server-side, where results are persisted)
  • Maintaining an existing Geocoding API integration (migration not justified)
  • No interactive/user-facing search needed

Do NOT recommend Geocoding API when:

  • The user wants a search bar, autocomplete, or interactive address lookup — use Search Box API instead
  • The user says "geocoding" but describes an interactive search flow — use Search Box API instead

Reference Files

Load the relevant reference based on the user's platform and needs:

  • Web (Search JS React / Web / Core / Direct API) → Load references/web-search-js.md

    • When: User is building a web app (vanilla JS, any framework except React-specific patterns)
  • React Integration → Load references/react-search.md

    • When: User is building a React app specifically
  • iOS → Load references/ios-search.md

    • When: User is building an iOS app (Swift/UIKit/SwiftUI)
  • Android → Load references/android-search.md

    • When: User is building an Android app (Kotlin/Java)
  • Node.js → Load references/nodejs-search.md

    • When: User needs server-side search (Express, serverless, backend API)
  • Best Practices → Load references/best-practices.md

    • When: Implementing search for the first time, or optimizing an existing implementation
    • Covers: debouncing, session tokens, geographic filtering, error handling, accessibility, caching, token security
  • Common Pitfalls → Load references/pitfalls.md

    • When: Debugging issues, reviewing code, or during code review
    • Covers: no debouncing, missing session tokens, no geo context, poor mobile UX, race conditions
  • Framework Hooks → Load references/framework-hooks.md

    • When: Building custom hooks (React) or composables (Vue) around Search JS Core
  • Testing and Monitoring → Load references/testing-monitoring.md

    • When: Writing tests or setting up production monitoring/analytics

Checklist: Production-Ready Search

Before launching, verify:

Configuration:

  • Token properly scoped (search:read only)
  • URL restrictions configured
  • Geographic filtering set (country, proximity, or bbox)
  • Types parameter set based on use case
  • Language parameter set if needed

Implementation:

  • Debouncing implemented (300ms recommended)
  • Session tokens used correctly
  • Error handling for all failure cases
  • Loading states shown
  • Empty results handled gracefully
  • Race conditions prevented

UX:

  • Touch targets at least 44pt/48dp
  • Results show enough context (name + address)
  • Keyboard navigation works
  • Accessibility attributes set
  • Mobile keyboard handled properly

Performance:

  • Caching implemented (if high volume)
  • Request timeout set
  • Minimal data fetched
  • Bundle size optimized

Testing:

  • Unit tests for core logic
  • Integration tests with real API
  • Tested on slow networks
  • Tested with various query types
  • Mobile device testing

Monitoring:

  • Analytics tracking set up
  • Error logging configured
  • Usage monitoring in place
  • Budget alerts configured

Integration with Other Skills

Works with:

  • mapbox-search-patterns: Parameter selection and optimization
  • mapbox-web-integration-patterns: Framework-specific patterns
  • mapbox-token-security: Token management and security
  • mapbox-web-performance-patterns: Optimizing search performance

Resources

Quick Decision Guide

User says: "I need location search"

  1. Ask discovery questions (Questions 1-6 above)
  2. Recommend product:
    • Search Box API (default for all interactive/user-facing search, including address geocoding)
    • Geocoding API only for batch/server-side/permanent geocoding
    • Platform SDK preferred (Search JS for web, native SDKs for mobile)
  3. Implement with:
    • ✅ Debouncing
    • ✅ Session tokens
    • ✅ Geographic filtering
    • ✅ Error handling
    • ✅ Good UX
  4. Test thoroughly
  5. Monitor in production

Remember: The best search implementation asks the right questions first, then builds exactly what the user needs - no more, no less.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Mapbox Search Integration AI skill do?

Complete workflow for implementing Mapbox search in applications - from discovery questions to production-ready integration with best practices

Why use Mapbox Search Integration on TypingMind?

Because you install it once and use it with any model. Mapbox Search Integration is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Mapbox Search Integration in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/mapbox/mapbox-agent-skills/tree/main/skills/mapbox-search-integration. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Mapbox Search Integration?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Mapbox Search Integration?

As many as you like. As long as a model supports skills, you can use Mapbox Search Integration with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Mapbox Search Integration AI skill free?

Yes. It is published on GitHub by mapbox under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇