Web Search logo

Web Search

Organization
brave
web-search

USE FOR web search. Returns ranked results with snippets, URLs, thumbnails. Supports freshness filters, SafeSearch, Goggles for custom ranking, pagination. Primary search endpoint.

Overview

Publisherbrave
Repositorybrave-search-skills
Skill nameweb-search
Stars
182
Forks
17
Bundled files
Instructions only
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.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

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

Installation

Install the Web 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/brave/brave-search-skills/tree/main/skills/web-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/brave/brave-search-skills.git /tmp/brave-search-skills
mkdir -p .claude/skills
cp -r /tmp/brave-search-skills/skills/web-search .claude/skills/web-search
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Web 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 Web 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 Web 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.

Web Search

Requires API Key: Get one at https://api.search.brave.com

Plan: Included in the Search plan. See https://api-dashboard.search.brave.com/app/subscriptions/subscribe

Quick Start (cURL)

Basic Search

bash
curl -s "https://api.search.brave.com/res/v1/web/search?q=python+web+frameworks" \
  -H "Accept: application/json" \
  -H "X-Subscription-Token: ${BRAVE_SEARCH_API_KEY}"

With Parameters

bash
curl -s "https://api.search.brave.com/res/v1/web/search" \
  -H "Accept: application/json" \
  -H "X-Subscription-Token: ${BRAVE_SEARCH_API_KEY}" \
  -G \
  --data-urlencode "q=rust programming tutorials" \
  --data-urlencode "country=US" \
  --data-urlencode "search_lang=en" \
  --data-urlencode "count=10" \
  --data-urlencode "safesearch=moderate" \
  --data-urlencode "freshness=pm"

Endpoint

http
GET https://api.search.brave.com/res/v1/web/search
POST https://api.search.brave.com/res/v1/web/search

Note: Both GET and POST methods are supported. POST is useful for long queries or complex Goggles.

Authentication: X-Subscription-Token: <API_KEY> header

Optional Headers:

  • Accept-Encoding: gzip — Enable gzip compression

When to Use Web Search

FeatureWeb Search (this)LLM Context (llm-context)Answers (answers)
OutputStructured results (links, snippets, metadata)Pre-extracted page content for LLMsEnd-to-end AI answers with citations
Result typesWeb, news, videos, discussions, FAQ, infobox, locations, richExtracted text chunks, tables, codeSynthesized answer + source list
Unique featuresGoggles, structured data (schemas), rich callbacksToken budget control, threshold modesMulti-iteration search, streaming, OpenAI SDK compatible
SpeedFast (~0.5-1s)Fast (<1s)Slower (~30-180s)
Best forSearch UIs, data extraction, custom rankingRAG pipelines, AI agents, groundingChat interfaces, thorough research

Parameters

ParameterTypeRequiredDefaultDescription
qstringYes-Search query (1-400 chars, max 50 words)
countrystringNoUSSearch country (2-letter country code or ALL)
search_langstringNoenLanguage preference (2+ char language code)
ui_langstringNoen-USUI language (e.g., "en-US")
countintNo20Max results per page (1-20)
offsetintNo0Page offset for pagination (0-9)
safesearchstringNomoderateAdult content filter (off/moderate/strict)
freshnessstringNo-Time filter (pd/pw/pm/py or date range)
text_decorationsboolNotrueInclude highlight markers
spellcheckboolNotrueAuto-correct query
result_filterstringNo-Filter result types (comma-separated)
gogglesstringNo-Custom ranking filter (URL or inline)
extra_snippetsboolNo-Get up to 5 extra snippets per result
operatorsboolNotrueApply search operators
unitsstringNo-Measurement units (metric/imperial)
enable_rich_callbackboolNofalseEnable rich 3rd party data callback
include_fetch_metadataboolNofalseInclude fetched_content_timestamp on results

Freshness Values

ValueDescription
pdPast day (24 hours)
pwPast week (7 days)
pmPast month (31 days)
pyPast year (365 days)
YYYY-MM-DDtoYYYY-MM-DDCustom date range

Result Filter Values

Filter types: discussions, faq, infobox, news, query, videos, web, locations

bash
# Only web and video results
curl "...&result_filter=web,videos"

Location Headers (Optional)

For location-aware results, add these headers. Lat/Long is sufficient when coordinates are known — the other headers are only needed as a fallback when coordinates are unavailable.

HeaderTypeDescription
X-Loc-LatfloatUser latitude (-90.0 to 90.0)
X-Loc-LongfloatUser longitude (-180.0 to 180.0)
X-Loc-TimezonestringIANA timezone (e.g., "America/San_Francisco")
X-Loc-CitystringCity name
X-Loc-StatestringState/region code (ISO 3166-2)
X-Loc-State-NamestringState/region full name (e.g., "California")
X-Loc-Countrystring2-letter country code
X-Loc-Postal-CodestringPostal code (e.g., "94105")

Priority: X-Loc-Lat + X-Loc-Long take precedence. When provided, downstream services resolve the location directly from coordinates and the text-based headers (City, State, Country, Postal-Code) are not used for location resolution. Provide text-based headers only when you don't have coordinates. Sending both won't break anything — lat/long simply wins.

Response Format

Response Fields

FieldTypeDescription
typestringAlways "search"
query.originalstringThe original search query
query.alteredstring?Spellcheck-corrected query (if changed)
query.cleanedstring?Cleaned/normalized query
query.spellcheck_offbool?Whether spellcheck was disabled
query.more_results_availableboolWhether more pages exist
query.show_strict_warningbool?True if strict safesearch blocked adult results
query.search_operatorsobject?Applied search operators (applied, cleaned_query, sites)
web.typestringAlways "search"
web.results[].titlestringPage title
web.results[].urlstringPage URL
web.results[].descriptionstring?Snippet/description text
web.results[].agestring?Human-readable age (e.g., "2 days ago")
web.results[].languagestring?Content language code
web.results[].meta_urlobjectURL components (scheme, netloc, hostname, path)
web.results[].thumbnailobject?Thumbnail (src, original)
web.results[].thumbnail.originalstring?Original full-size image URL
web.results[].thumbnail.logobool?Whether the thumbnail is a logo
web.results[].profileobject?Publisher identity (name, url, long_name, img)
web.results[].page_agestring?ISO datetime of publication (e.g., "2025-04-12T14:22:41")
web.results[].extra_snippetslist[str]?Up to 5 additional excerpts
web.results[].deep_resultsobject?Additional links (buttons, links) from the page
web.results[].schemaslist?Raw schema.org structured data
web.results[].productobject?Product info and reviews
web.results[].recipeobject?Recipe details (ingredients, time, ratings)
web.results[].articleobject?Article metadata (author, publisher, date)
web.results[].bookobject?Book info (author, ISBN, rating)
web.results[].softwareobject?Software product info
web.results[].ratingobject?Aggregate ratings
web.results[].faqobject?FAQ found on the page
web.results[].movieobject?Movie info (directors, actors, genre)
web.results[].videoobject?Video metadata (duration, views, creator)
web.results[].locationobject?Location/restaurant details
web.results[].qaobject?Question/answer info
web.results[].creative_workobject?Creative work data
web.results[].music_recordingobject?Music/song data
web.results[].organizationobject?Organization info
web.results[].reviewobject?Review data
web.results[].content_typestring?Content type classification
web.results[].fetched_content_timestampint?Fetch timestamp (with include_fetch_metadata=true)
web.mutated_by_gogglesboolWhether results were re-ranked by Goggles
web.family_friendlyboolWhether results are family-friendly
mixedobject?Preferred display order (see Mixed Response below)
discussions.results[]array?Forum discussion clusters
discussions.results[].data.forum_namestring?Forum/community name
discussions.results[].data.num_answersint?Number of answers/replies
discussions.results[].data.questionstring?Discussion question
discussions.results[].data.top_commentstring?Top-voted comment excerpt
faq.results[]array?FAQ entries
news.results[]array?News articles
videos.results[]array?Video results
infobox.results[]array?Knowledge graph entries
locations.results[]array?Local POI results
rich.hint.verticalstring?Rich result type
rich.hint.callback_keystring?Callback key for rich data

JSON Example

json
{
  "type": "search",
  "query": {
    "original": "python frameworks",
    "altered": "python web frameworks",
    "spellcheck_off": false,
    "more_results_available": true
  },
  "web": {
    "type": "search",
    "results": [
      {
        "title": "Top Python Web Frameworks",
        "url": "https://example.com/python-frameworks",
        "description": "A comprehensive guide to Python web frameworks...",
        "age": "2 days ago",
        "language": "en",
        "meta_url": {
          "scheme": "https",
          "netloc": "example.com",
          "hostname": "example.com",
          "path": "/python-frameworks"
        },
        "thumbnail": {
          "src": "https://...",
          "original": "https://original-image-url.com/img.jpg"
        },
        "extra_snippets": ["Additional excerpt 1...", "Additional excerpt 2..."]
      }
    ],
    "family_friendly": true
  },
  "mixed": {
    "type": "mixed",
    "main": [
      {"type": "web", "index": 0, "all": false},
      {"type": "web", "index": 1, "all": false},
      {"type": "videos", "all": true}
    ],
    "top": [],
    "side": []
  },
  "videos": { "...": "..." },
  "news": { "...": "..." },
  "rich": {
    "type": "rich",
    "hint": {
      "vertical": "weather",
      "callback_key": "<callback_key_hex>"
    }
  }
}

Mixed Response

The mixed object defines the preferred display order of results across types. It contains three arrays:

ArrayPurpose
mainPrimary result list (ordered sequence of results to display)
topResults to display above main results
sideResults to display alongside main results (e.g., infobox)

Each entry is a ResultReference with type (e.g., "web", "videos"), index (into the corresponding result array), and all (true to include all results of that type at this position).

Search Operators

OperatorSyntaxDescription
Sitesite:example.comLimit results to a specific domain
File extensionext:pdfResults with a specific file extension
File typefiletype:pdfResults created in a specific file type
In titleintitle:pythonPages with term in the title
In bodyinbody:tutorialPages with term in the body
In pageinpage:guidePages with term in title or body
Languagelang:esPages in a specific language (ISO 639-1)
Locationloc:usPages from a specific country (ISO 3166-1 alpha-2)
Include+termForce inclusion of a term
Exclude-termExclude pages containing the term
Exact match"exact phrase"Match the exact phrase in order
ANDterm1 AND term2Both terms required (uppercase)
OR / NOTterm1 OR term2, NOT termLogical operators (uppercase)

Set operators=false to disable operator parsing.

Goggles (Custom Ranking) — Unique to Brave

Goggles let you re-rank search results — boost trusted sources, suppress SEO spam, or build focused search scopes.

MethodExample
Hosted--data-urlencode "goggles=https://raw.githubusercontent.com/brave/goggles-quickstart/main/goggles/rust_programming.goggle"
Inline--data-urlencode 'goggles=$discard\n$site=example.com'

Hosted goggles must be on GitHub/GitLab, include ! name:, ! description:, ! author: headers, and be registered at https://search.brave.com/goggles/create. Inline rules need no registration.

Syntax: Rules start with $ + comma-separated options. Actions (pick one): discard, boost[=N], downrank[=N] — N is an integer 1–10. Site filter: site=DOMAIN. Example: $site=example.com,boost=3. Separate rules with \n (%0A).

Allow list: $discard\n$site=docs.python.org\n$site=developer.mozilla.org — Block list: $discard,site=pinterest.com\n$discard,site=quora.com

Resources: Discover · Syntax · Quickstart

Rich Data Enrichments

For queries about weather, stocks, sports, currency, etc., use the rich callback workflow:

bash
# 1. Search with rich callback enabled
curl -s "https://api.search.brave.com/res/v1/web/search?q=weather+san+francisco&enable_rich_callback=true" \
  -H "X-Subscription-Token: ${BRAVE_SEARCH_API_KEY}"

# Response includes: "rich": {"hint": {"callback_key": "abc123...", "vertical": "weather"}}

# 2. Get rich data with the callback key
curl -s "https://api.search.brave.com/res/v1/web/rich?callback_key=abc123..." \
  -H "X-Subscription-Token: ${BRAVE_SEARCH_API_KEY}"

Supported Rich Types: Calculator, Definitions, Unit Conversion, Unix Timestamp, Package Tracker, Stock, Currency, Cryptocurrency, Weather, American Football, Baseball, Basketball, Cricket, Football/Soccer, Ice Hockey, Web3, Translator

Rich Callback Endpoint

http
GET https://api.search.brave.com/res/v1/web/rich
ParameterTypeRequiredDescription
callback_keystringYesCallback key from the web search rich.hint.callback_key field

Use Cases

  • General-purpose search integration: Richest result set (web, news, videos, discussions, FAQ, infobox, locations) in one call. For RAG/LLM grounding, prefer llm-context.
  • Structured data extraction: Products, recipes, ratings, articles via schemas and typed fields on results.
  • Custom search with Goggles: Unique to Brave. Boost/discard sites with inline rules or hosted Goggles for fully customized ranking.

Notes

  • Pagination: Use offset (0-9) with count to page through results
  • Count: Max 20 for web search; actual results may be less than requested

Frequently asked questions

What does the Web Search AI skill do?

USE FOR web search. Returns ranked results with snippets, URLs, thumbnails. Supports freshness filters, SafeSearch, Goggles for custom ranking, pagination. Primary search endpoint.

Why use Web Search on TypingMind?

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

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

Which AI models can use Web 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 Web Search?

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

Is the Web Search AI skill free?

Yes. It is published on GitHub by brave 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 👇