Places Search logo

Places Search

Community
win4r
places_search

Find, compare, and share details on physical places near the user or in a specified area, including restaurants, cafes, bars, hotels, parks, attractions, shops, and businesses with local services. Not for itineraries, choosing a city or region, dated events or showtimes, or directions.

Overview

Publisherwin4r
RepositoryMuseAI-Skills
Skill nameplaces_search
Stars
382
Forks
116
Bundled files
1
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.

  • 1 bundled files

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

  • Open source

    Published by win4r on GitHub. Read the source before you install it.

Installation

Install the Places Search 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.
    https://github.com/win4r/MuseAI-Skills/tree/main/opt/hatch/skills/places-search
  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/win4r/MuseAI-Skills.git /tmp/MuseAI-Skills
mkdir -p .claude/skills
cp -r /tmp/MuseAI-Skills/opt/hatch/skills/places-search .claude/skills/places_search
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Places Search 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 Places Search 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 Places Search 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.

Places

Use browser.search for all place discovery and name/address lookup. Run places <command> through muse.exec for richer details and photo-based place detection. Never load or search for a places tool namespace. If a search comes back empty or weak, refine the browser.search query per the rules below. Print CLI JSON to stdout for you to read, not to show the user. Use --help when needed. No sign-in is required.

Common flows

Find places

Use browser.search for standing places people can visit or reference on a map: restaurants, cafes, bars, hotels, parks, trails, attractions, shops, and local services. A follow-up like "anything cheaper?" or "Italian instead?" refines the preceding search.

  • Put the complete user intent in primary_query.query. Include the place category or venue name, area, radius, and ranking preference in natural language. For example, search for "best sushi restaurants in San Francisco within five miles", not just "sushi".
  • Preserve "near me" or "nearby" in the query when that is what the user asked so that device location is used; never invent or guess a location.
  • One call covers one area. When the user names multiple cities or neighborhoods, run a separate call for each and cover each in the answer.
  • For category discovery, include the category and area in the query. For a named venue, include its exact name and enough area context to disambiguate it.
  • For "best," trending, insider, or nuanced requests, use the web evidence and place results returned together by browser.search. Ground every named venue in a returned result.
  • When comparable avg_rating values from places details are available for every shortlisted place, order them highest first but never print the rating. Otherwise order by relevance to the user's constraints.
  • Avoid showing permanently or temporarily closed places unless specifically asked for.
  • Judge the returned result set rather than following rank blindly. Set aside wrong-category or wrong-area results. If the set is weak, retry with an exact venue name, a narrower subtype, or clearer area language.

Get richer details

  • places details takes numeric place IDs only. It cannot search by name or address; use browser.search for that.
  • Call places details only when browser.search explicitly returns an all-digit place_id for that result. Never derive an ID from a venue name or URL.
  • Prefer the browser.search response when it already contains enough detail. Use places details for richer or fresher hours, prices, photos, reviews, and offerings.
  • If browser.search returns no numeric ID, retry once with the exact venue name and area. If that still returns no ID, answer from cited search evidence without details or a map.
  • Batch several IDs into one call when comparing places. Pass --motivation with the user's intent so quotes and photos are ranked for it.
  • places details returns {"<place_id>": {"details": {…, "rating": {"avg_rating", "num_rating"}}}}, keyed by id.

Show places on a map

Before creating a map, run places details for the selected numeric IDs. Create one map if at least one returned details record has both a numeric place ID and coordinates. This applies to recommendations, comparisons, named-place lookups, and other uses where a map could ground the user. See Response formatting for which places go on it.

Create one local_map widget with widget.create and this payload:

json
{
  "kind": "local_map",
  "data": {
    "elements": [
      {
        "kind": "rich_place",
        "place_id": "<numeric ID>",
        "name": "<name from the same result>",
        "coordinate": {
          "latitude": 0.0,
          "longitude": 0.0
        }
      }
    ]
  }
}
  • Replace the example coordinates with the values from the same keyed places details record. Use its canonical name. Omit a place whose details or coordinates are missing; never infer an ID or geocode a name.
  • If creation fails, answer in text. Do not retry or build an HTML or image map.

Identify a place from a photo

Use places detect when captured frames and GPS need to resolve which place the user is at. It returns ranked candidates as {place_id, place_name, confidence}. Pass a returned ID to places details for more. The optional 4th --location field is the connected Wi-Fi BSSID; include it verbatim when known to improve Home/Work matching, and never echo a raw BSSID back to the user.

Response formatting

  • Never name a place that did not come back from a tool.
  • Match the framing to the ask. When the user wants a recommendation, give them two to three. When the request is broad, offer a few named results as a starting point.
  • Name two to three places in text from the merged result set. Write each on its own line in exactly this shape: **Name**, a sentence or two on why you chose it. Open with one short line framing the answer before the places.
  • Send the text as one message. Never add a second message after the map.
  • No addresses, raw place IDs, coordinates, star ratings, review counts, or price levels in text.
  • After a recommendation, note there are more options without itemizing; after a broad answer, offer to narrow to a recommendation.
  • Cite factual claims sourced from browser.search using its normal citation format.
  • Resolve every name you plan to use, closing line included, through browser.search before creating the map. Never name an unresolved place.
  • Build the map from the most relevant places returned that have a place ID and coordinates. Don't rely solely on named places in your text as an indication of what places should appear on the map.
  • The places you named come first on the map, in the order you named them, ahead of everything else.
  • Create one map per answer. Include the returned embed_token.
  • Never narrate or reference the map. No "I mapped things out for you"; "Take a look at the map", etc.

Limits

  • Do not write directions, turn-by-turn steps, maps links, or navigation controls.
  • Do not estimate travel times or distances. Use only values returned by a tool.

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 Places Search AI skill do?

Find, compare, and share details on physical places near the user or in a specified area, including restaurants, cafes, bars, hotels, parks, attractions, shops, and businesses with local services. Not for itineraries, choosing a city or region, dated events or showtimes, or directions.

Why use Places Search on TypingMind?

Because you install it once and use it with any model. Places Search 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 Places Search in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/win4r/MuseAI-Skills/tree/main/opt/hatch/skills/places-search. 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 Places Search?

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 Places Search?

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

Is the Places Search AI skill free?

It is published on GitHub by win4r. Check the repository for licensing terms. 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 👇