Brave Search

Connect to Brave Search to run privacy-first web, news, and local searches, retrieve AI summaries, get LLM grounding context, and use search-backed chat completions.

Connect to Brave Search to run privacy-first web, news, image, video, and local searches; retrieve AI-generated summaries and entity data; get LLM-optimized context for grounding; and use OpenAI-compatible chat completions backed by real-time search results.

Supports authentication: API Key

What you can build with this connector

Use case	Tools involved
Real-time web research	`brave_web_search` → feed results into LLM context
News monitoring	`brave_news_search` with `freshness: pd` → summarise headlines
LLM grounding	`brave_llm_context` → pass structured snippets directly to your model
Search-augmented chat	`brave_chat_completions` with conversation history → cited AI answers
Local business lookup	`brave_local_place_search` → `brave_local_pois` → display hours, ratings, address
AI summaries with follow-ups	`brave_web_search` (summary: true) → `brave_summarizer_search` → `brave_summarizer_followups`

Key concepts:

Plan tiers: Free covers core search. Pro adds brave_llm_context and summarizer tools. AI plan adds brave_chat_completions. Data for AI plan adds local/POI tools.
Summarizer two-step: First call brave_web_search with summary: true to get a summarizer.key, then pass that key to brave_summarizer_* tools.
LLM context vs web search: brave_llm_context returns token-budgeted, snippet-optimised output specifically for grounding — prefer it over raw web search results when feeding an LLM.
Rate limits: Free plan allows 1 req/s. Paid plans scale to 20 req/s. Handle 429 errors with backoff.

Set up the agent connector

Register your Brave Search API key with Scalekit so it can authenticate and proxy search requests on behalf of your users. Unlike OAuth connectors, Brave Search uses API key authentication — there is no redirect URI or OAuth flow.

Get a Brave Search API key

Go to api.search.brave.com and sign in or create a free account.
In the left sidebar, click API Keys → + New Key. Give it a name (e.g., Agent Auth) and click Create.
Copy the key immediately — it is shown only once.

Brave Search API dashboard showing API Keys page with existing keys and the New Key button

Brave Search API offers several subscription tiers. Make sure your plan covers the tools you intend to use:

Plan	Monthly free quota	Tools included
Free	2,000 queries/month, 1 req/s	`brave_web_search`, `brave_news_search`, `brave_image_search`, `brave_video_search`, `brave_suggest_search`, `brave_spellcheck`
Base / Pro	Pay-per-use ($3–$5 / 1,000 queries)	All Free tools + `brave_llm_context`, `brave_summarizer_*` (Pro and above)
AI	Pay-per-use	All Pro tools + `brave_chat_completions`
Data for AI	Pay-per-use	Adds `brave_local_place_search`, `brave_local_pois`, `brave_local_descriptions`

Upgrade your plan at api.search.brave.com → Subscription.

Create a connection in Scalekit
- In Scalekit dashboard, go to Agent Auth → Create Connection. Find Brave Search and click Create.
- Note the Connection name — you will use this as connection_name in your code (e.g., brave-search).
Add a connected account

Connected accounts link a specific user identifier in your system to a Brave Search API key. Add them via the dashboard for testing, or via the Scalekit API in production.

Via dashboard (for testing)
- Open the connection you created and click the Connected Accounts tab → Add account.
- Fill in:
  - Your User’s ID — a unique identifier for this user in your system (e.g., user_123)
  - API Key — the Brave Search API key you copied in step 1
- Click Save.
Via API (for production)
- Node.js
- Python
1 await scalekit.actions.upsertConnectedAccount({ 2 connectionName: 'brave-search', 3 identifier: 'user_123', // your user's unique ID 4 credentials: { api_key: 'BSA...' }, 5 });
1 scalekit_client.actions.upsert_connected_account( 2 connection_name="brave-search", 3 identifier="user_123", 4 credentials={"api_key": "BSA..."} 5 )
In production, call upsert_connected_account (Python) / upsertConnectedAccount (Node.js) when a user enters their Brave Search API key — for example, on a settings page in your app.

Usage

Once a connected account is set up, make search API calls through the Scalekit proxy. Scalekit injects the Brave Search API key automatically as the X-Subscription-Token header — you never handle credentials in your application code.

You can interact with Brave Search in two ways — via direct proxy API calls or via Scalekit optimized tool calls. Scroll down to see the list of available Scalekit tools.

1
import { ScalekitClient } from '@scalekit-sdk/node';
2
import 'dotenv/config';
3

4
const connectionName = 'brave-search'; // connection name from your Scalekit dashboard
5
const identifier = 'user_123';         // your user's unique identifier
6

7
// Get your credentials from app.scalekit.com → Developers → Settings → API Credentials
8
const scalekit = new ScalekitClient(
9
  process.env.SCALEKIT_ENV_URL,
10
  process.env.SCALEKIT_CLIENT_ID,
11
  process.env.SCALEKIT_CLIENT_SECRET
12
);
13
const actions = scalekit.actions;
14

15
// Web search via Scalekit proxy — no API key needed here
16
const result = await actions.request({
17
  connectionName,
18
  identifier,
19
  path: '/res/v1/web/search',
20
  method: 'GET',
21
  queryParams: { q: 'best open source LLM frameworks 2025', count: '5' },
22
});
23
console.log(result.data.web.results);

1
import scalekit.client, os
2
from dotenv import load_dotenv
3
load_dotenv()
4

5
connection_name = "brave-search"  # connection name from your Scalekit dashboard
6
identifier = "user_123"           # your user's unique identifier
7

8
# Get your credentials from app.scalekit.com → Developers → Settings → API Credentials
9
scalekit_client = scalekit.client.ScalekitClient(
10
    client_id=os.getenv("SCALEKIT_CLIENT_ID"),
11
    client_secret=os.getenv("SCALEKIT_CLIENT_SECRET"),
12
    env_url=os.getenv("SCALEKIT_ENV_URL"),
13
)
14
actions = scalekit_client.actions
15

16
# Web search via Scalekit proxy — no API key needed here
17
result = actions.request(
18
    connection_name=connection_name,
19
    identifier=identifier,
20
    path="/res/v1/web/search",
21
    method="GET",
22
    params={"q": "best open source LLM frameworks 2025", "count": 5}
23
)
24
print(result["web"]["results"])

Scalekit tools

Use actions.execute_tool() to call Brave Search tools directly. Scalekit injects credentials automatically — your application code never handles the API key.

Web search

Search the web and retrieve real-time results. Works on all plans including Free.

1
import os
2
from scalekit.client import ScalekitClient
3

4
scalekit_client = ScalekitClient(
5
    client_id=os.environ["SCALEKIT_CLIENT_ID"],
6
    client_secret=os.environ["SCALEKIT_CLIENT_SECRET"],
7
    env_url=os.environ["SCALEKIT_ENV_URL"],
8
)
9

10
# Ensure a connected account exists for this user before making tool calls
11
scalekit_client.actions.get_or_create_connected_account(
12
    connection_name="brave-search",
13
    identifier="user_123",
14
)
15

16
# Search for recent articles — freshness "pw" limits results to the past 7 days
17
result = scalekit_client.actions.execute_tool(
18
    connection_name="brave-search",
19
    identifier="user_123",
20
    tool_name="brave_web_search",
21
    input={
22
        "q": "open source LLM frameworks 2025",
23
        "count": 5,
24
        "freshness": "pw",
25
    },
26
)
27

28
for item in result["web"]["results"]:
29
    print(item["title"], item["url"])

News search

Retrieve recent news articles by topic or keyword. Useful for monitoring a brand, topic, or competitor.

1
# Fetch the latest news on a topic from the past 24 hours
2
result = scalekit_client.actions.execute_tool(
3
    connection_name="brave-search",
4
    identifier="user_123",
5
    tool_name="brave_news_search",
6
    input={
7
        "q": "AI regulation Europe",
8
        "count": 10,
9
        "freshness": "pd",   # pd = past 24 hours
10
    },
11
)
12

13
for article in result["results"]:
14
    print(article["title"], article["age"], article["url"])

LLM grounding context

Retrieve search results structured specifically for grounding LLM responses. Requires Pro plan. Use token_budget to stay within your model’s context window.

1
# Get search context sized to fit a 4 000-token budget
2
result = scalekit_client.actions.execute_tool(
3
    connection_name="brave-search",
4
    identifier="user_123",
5
    tool_name="brave_llm_context",
6
    input={
7
        "q": "vector database comparison 2025",
8
        "count": 5,
9
        "token_budget": 4000,
10
    },
11
)
12

13
# Pass the structured context directly to your LLM
14
grounding_context = result["context"]
15
print(grounding_context)

AI chat completions grounded in search

Get an AI-generated answer backed by real-time Brave Search results, using an OpenAI-compatible interface. Requires AI plan.

1
# Drop-in replacement for OpenAI /v1/chat/completions — results are grounded in live search
2
result = scalekit_client.actions.execute_tool(
3
    connection_name="brave-search",
4
    identifier="user_123",
5
    tool_name="brave_chat_completions",
6
    input={
7
        "messages": [
8
            {"role": "user", "content": "What are the latest developments in quantum computing?"}
9
        ],
10
        "model": "brave/serp-claude-3-5-haiku",
11
    },
12
)
13

14
print(result["choices"][0]["message"]["content"])
15
# Each answer includes citations back to source URLs
16
for source in result.get("search_results", []):
17
    print(source["title"], source["url"])

Local place search and POI details

Find nearby businesses or points of interest, then retrieve full details. Requires Data for AI plan.

1
# Step 1: Find coffee shops near San Francisco city centre
2
places = scalekit_client.actions.execute_tool(
3
    connection_name="brave-search",
4
    identifier="user_123",
5
    tool_name="brave_local_place_search",
6
    input={
7
        "q": "specialty coffee",
8
        "location": "San Francisco, CA",
9
        "count": 5,
10
    },
11
)
12

13
location_ids = [p["id"] for p in places["results"]]
14

15
# Step 2: Fetch rich details (hours, ratings, address) for those locations
16
# Location IDs expire after ~8 hours — always fetch details in the same session
17
pois = scalekit_client.actions.execute_tool(
18
    connection_name="brave-search",
19
    identifier="user_123",
20
    tool_name="brave_local_pois",
21
    input={"ids": location_ids},
22
)
23

24
for poi in pois["results"]:
25
    print(poi["name"], poi["address"], poi["rating"])

AI summary with follow-up queries

Get an AI-generated summary for a search query, then surface follow-up questions. Requires Pro plan.

1
# Step 1: Web search with summary: true to obtain a summarizer key
2
search_result = scalekit_client.actions.execute_tool(
3
    connection_name="brave-search",
4
    identifier="user_123",
5
    tool_name="brave_web_search",
6
    input={
7
        "q": "benefits of RAG vs fine-tuning for enterprise LLMs",
8
        "summary": True,
9
        "count": 5,
10
    },
11
)
12

13
summarizer_key = search_result["summarizer"]["key"]
14

15
# Step 2: Retrieve the full AI summary using the key
16
summary = scalekit_client.actions.execute_tool(
17
    connection_name="brave-search",
18
    identifier="user_123",
19
    tool_name="brave_summarizer_search",
20
    input={"key": summarizer_key, "entity_info": True},
21
)
22

23
print(summary["title"])
24
print(summary["summary"])
25

26
# Step 3: Get follow-up questions for a conversational search experience
27
followups = scalekit_client.actions.execute_tool(
28
    connection_name="brave-search",
29
    identifier="user_123",
30
    tool_name="brave_summarizer_followups",
31
    input={"key": summarizer_key},
32
)
33

34
for q in followups["queries"]:
35
    print("-", q)

LangChain integration

Use Scalekit’s LangChain helper to load all Brave Search tools into a LangChain agent. The agent selects and calls the right tool automatically based on the user’s query.

1
import os
2
from langchain.agents import AgentExecutor, create_tool_calling_agent
3
from langchain_anthropic import ChatAnthropic
4
from langchain_core.prompts import ChatPromptTemplate
5
from scalekit.client import ScalekitClient
6

7
scalekit_client = ScalekitClient(
8
    client_id=os.environ["SCALEKIT_CLIENT_ID"],
9
    client_secret=os.environ["SCALEKIT_CLIENT_SECRET"],
10
    env_url=os.environ["SCALEKIT_ENV_URL"],
11
)
12

13
# Load all Brave Search tools — Scalekit handles auth for each call
14
tools = scalekit_client.actions.langchain.get_tools(
15
    providers=["BRAVE_SEARCH"],
16
    identifier="user_123",
17
)
18

19
llm = ChatAnthropic(
20
    model="claude-sonnet-4-6",
21
    api_key=os.environ["ANTHROPIC_API_KEY"],
22
)
23

24
prompt = ChatPromptTemplate.from_messages([
25
    ("system", "You are a helpful research assistant with access to Brave Search tools."),
26
    ("human", "{input}"),
27
    ("placeholder", "{agent_scratchpad}"),
28
])
29

30
agent = create_tool_calling_agent(llm, tools, prompt)
31
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)
32

33
response = agent_executor.invoke({
34
    "input": (
35
        "Find the top 5 news articles about AI regulation in Europe from the past week "
36
        "and give me a one-sentence summary of each."
37
    )
38
})
39
print(response["output"])

Tool list

The following tools are available when you connect a Brave Search account. Each tool maps to a Brave Search API endpoint. Required plan is noted per tool.

`brave_web_search`

Search the web using Brave’s privacy-focused search engine. Returns real-time results including web pages, news, videos, images, and rich data. Supports filtering by country, language, recency, and custom re-ranking via Goggles.

Required plan: Free and above.

Name	Type	Required	Description
`q`	string	Yes	Search query. Maximum 400 characters, 50 words.
`country`	string	No	Country code (ISO 3166-1 alpha-2, e.g., `US`, `GB`, `DE`) to localise results.
`search_lang`	string	No	Language of search results (ISO 639-1, e.g., `en`, `fr`, `de`).
`ui_lang`	string	No	UI language for rendering result labels (e.g., `en-US`).
`count`	integer	No	Number of results per page (1–20, default 20).
`offset`	integer	No	Pagination offset (0–9). Use with `count` to page through up to 200 results.
`safesearch`	string	No	Content filter: `off`, `moderate` (default), or `strict`.
`freshness`	string	No	Recency filter: `pd` (24 h), `pw` (7 days), `pm` (31 days), `py` (1 year), or a date range `YYYY-MM-DDtoYYYY-MM-DD`.
`text_decorations`	boolean	No	Include bold markers in result snippets for query-term highlighting.
`spellcheck`	boolean	No	Automatically spellcheck the query before searching.
`goggles_id`	string	No	URL of a Goggles re-ranking file for custom result ordering.
`units`	string	No	Unit system for unit-sensitive results: `metric` or `imperial`.
`extra_snippets`	boolean	No	Return up to 5 additional snippets per result. Requires Pro plan.
`summary`	boolean	No	Request an AI summarizer key in the response for use with `brave_summarizer_` tools. Requires Pro plan.*

`brave_news_search`

Search for recent news articles using Brave Search. Returns results with titles, URLs, snippets, publication dates, and source information. Supports filtering by country, language, and recency.

Required plan: Free and above.

Name	Type	Required	Description
`q`	string	Yes	News search query.
`country`	string	No	Country code (ISO 3166-1 alpha-2) to localise news results.
`search_lang`	string	No	Language of news results (ISO 639-1).
`count`	integer	No	Number of results (1–20, default 20).
`offset`	integer	No	Pagination offset (0–4).
`freshness`	string	No	Recency filter: `pd` (24 h), `pw` (7 days), `pm` (31 days).
`extra_snippets`	boolean	No	Return additional text snippets per result. Requires Pro plan.
`goggles_id`	string	No	URL of a Goggles re-ranking file.
`safesearch`	string	No	Content filter: `off`, `moderate` (default), or `strict`.
`spellcheck`	boolean	No	Automatically spellcheck the query before searching.

`brave_image_search`

Search for images using Brave Search. Returns image results with thumbnails, source URLs, dimensions, and metadata.

Required plan: Free and above.

Name	Type	Required	Description
`q`	string	Yes	Image search query.
`country`	string	No	Country code (ISO 3166-1 alpha-2) to localise results.
`search_lang`	string	No	Language filter (ISO 639-1).
`count`	integer	No	Number of image results to return (1–3 per API call).
`safesearch`	string	No	Content filter: `off`, `moderate` (default), or `strict`.
`spellcheck`	boolean	No	Automatically spellcheck the query before searching.

`brave_video_search`

Search for videos using Brave Search. Returns results with titles, URLs, thumbnails, durations, and publisher metadata.

Required plan: Free and above.

Name	Type	Required	Description
`q`	string	Yes	Video search query.
`country`	string	No	Country code (ISO 3166-1 alpha-2) to localise results.
`search_lang`	string	No	Language filter (ISO 639-1).
`count`	integer	No	Number of results (1–20, default 20).
`offset`	integer	No	Pagination offset (0–9).
`freshness`	string	No	Recency filter: `pd`, `pw`, `pm`, or a date range.
`safesearch`	string	No	Content filter: `off`, `moderate` (default), or `strict`.
`spellcheck`	boolean	No	Automatically spellcheck the query before searching.

`brave_suggest_search`

Get autocomplete search suggestions from Brave Search for a given query prefix. Useful for query completion, exploring related search terms, and building search UIs.

Required plan: Free and above.

Name	Type	Required	Description
`q`	string	Yes	Query prefix to get autocomplete suggestions for.
`country`	string	No	Country code (ISO 3166-1 alpha-2) to localise suggestions.
`count`	integer	No	Number of suggestions to return (1–20, default 5).

`brave_spellcheck`

Check and correct spelling of a query using Brave Search’s spellcheck engine. Returns suggested corrections for misspelled queries.

Required plan: Free and above.

Name	Type	Required	Description
`q`	string	Yes	The query to spellcheck.
`country`	string	No	Country code (ISO 3166-1 alpha-2) to apply locale-aware corrections.

`brave_llm_context`

Retrieve real-time web search results structured as grounding context for LLMs. Returns curated snippets, source URLs, titles, and metadata specifically formatted to maximise contextual relevance for AI-generated answers. Supports fine-grained token and snippet budgets to control context size.

Required plan: Pro and above.

Name	Type	Required	Description
`q`	string	Yes	Search query to retrieve grounding context for.
`count`	integer	No	Number of web results to include (1–20, default 5).
`token_budget`	integer	No	Maximum total tokens for the returned context. Use to fit within your LLM’s context window.
`snippet_budget`	integer	No	Maximum number of snippets to include across all results.
`country`	string	No	Country code (ISO 3166-1 alpha-2) to localise results.
`search_lang`	string	No	Language filter (ISO 639-1).
`freshness`	string	No	Recency filter: `pd`, `pw`, `pm`, `py`, or date range.
`safesearch`	string	No	Content filter: `off`, `moderate` (default), or `strict`.

`brave_chat_completions`

Get AI-generated answers grounded in real-time Brave Search results using an OpenAI-compatible chat completions interface. Returns summarised, cited answers with source references and token usage statistics. Drop-in replacement for OpenAI /v1/chat/completions for search-augmented generation.

Required plan: AI plan.

Name	Type	Required	Description
`messages`	array	Yes	Conversation history in OpenAI format: `[{"role": "user", "content": "..."}]`. Supported roles: `system`, `user`, `assistant`.
`model`	string	No	Model identifier (e.g., `brave/serp-claude-3-5-haiku`). Defaults to the plan’s included model.
`stream`	boolean	No	Stream the response using server-sent events (SSE). Default `false`.
`country`	string	No	Country code (ISO 3166-1 alpha-2) to localise the underlying search.
`search_lang`	string	No	Language filter for the underlying search (ISO 639-1).
`safesearch`	string	No	Content filter applied to the underlying search: `off`, `moderate`, or `strict`.
`freshness`	string	No	Recency filter for the underlying search results.

`brave_local_place_search`

Search 200M+ Points of Interest (POIs) by geographic centre and radius. Supports searching by coordinates or location name with an optional keyword query. Returns location IDs that you can use with brave_local_pois and brave_local_descriptions to get full details.

Required plan: Data for AI plan.

Name	Type	Required	Description
`q`	string	No	Keyword to filter POIs (e.g., `coffee shop`, `hospital`).
`lat`	number	No	Latitude of the search centre (–90 to 90). Use with `lon` instead of `location`.
`lon`	number	No	Longitude of the search centre (–180 to 180). Use with `lat` instead of `location`.
`location`	string	No	Human-readable location name (e.g., `San Francisco, CA`). Alternative to `lat`/`lon`.
`radius`	integer	No	Search radius in metres from the centre point.
`count`	integer	No	Number of POI results to return (1–20, default 5).

`brave_local_pois`

Fetch detailed Point of Interest data for up to 20 location IDs returned by brave_local_place_search. Returns rich local business data including address, phone, opening hours, ratings, and reviews.

Required plan: Data for AI plan.

Name	Type	Required	Description
`ids`	array	Yes	List of location IDs from a `brave_local_place_search` response (maximum 20 IDs per call).

`brave_local_descriptions`

Fetch AI-generated descriptions for locations using IDs from a brave_local_place_search response. Returns natural language summaries describing the place, its atmosphere, and what visitors can expect.

Required plan: Data for AI plan.

Name	Type	Required	Description
`ids`	array	Yes	List of location IDs from a `brave_local_place_search` response (maximum 20 IDs per call).

`brave_summarizer_search`

Retrieve a full AI-generated summary for a summarizer key obtained from a brave_web_search response (requires summary: true on the web search call). Returns the complete summary with title, content, enrichments, follow-up queries, and entity details.

Required plan: Pro and above.

Name	Type	Required	Description
`key`	string	Yes	Summarizer key from the `summarizer.key` field of a `brave_web_search` response.
`entity_info`	boolean	No	Include entity metadata (people, places, organisations) referenced in the summary.
`raw`	boolean	No	Return unformatted (raw) summary text without inline citation markers.

`brave_summarizer_summary`

Fetch only the complete AI-generated summary content for a summarizer key. Use when you only need the summary text without enrichments or follow-up data.

Required plan: Pro and above.

Name	Type	Required	Description
`key`	string	Yes	Summarizer key from a `brave_web_search` response.

`brave_summarizer_enrichments`

Fetch enrichment data for a Brave AI summary key. Returns associated images, Q&A pairs, entity details, and source references that accompany the summary.

Required plan: Pro and above.

Name	Type	Required	Description
`key`	string	Yes	Summarizer key from a `brave_web_search` response.

`brave_summarizer_followups`

Fetch suggested follow-up queries for a Brave AI summary key. Useful for building conversational search flows and helping users explore related topics.

Required plan: Pro and above.

Name	Type	Required	Description
`key`	string	Yes	Summarizer key from a `brave_web_search` response.

`brave_summarizer_entity_info`

Fetch detailed entity metadata for a specific entity mentioned in a Brave AI summary. Returns structured information about people, places, organisations, and concepts referenced in the summary.

Required plan: Pro and above.

Name	Type	Required	Description
`key`	string	Yes	Summarizer key from a `brave_web_search` response.
`entity`	string	Yes	Name of the entity to retrieve details for (must be present in the summary).

`brave_summarizer_title`

Fetch only the title component of a Brave AI summary. Use when you need a short heading for the summary without loading the full content.

Required plan: Pro and above.

Name	Type	Required	Description
`key`	string	Yes	Summarizer key from a `brave_web_search` response.

Brave Search

Set up the agent connector

Get a Brave Search API key

Create a connection in Scalekit

Add a connected account

Usage

Proxy API calls

Scalekit tools

Web search

News search

LLM grounding context

AI chat completions grounded in search

Local place search and POI details

AI summary with follow-up queries

LangChain integration

Tool list

brave_web_search

brave_news_search

brave_image_search

brave_video_search

brave_suggest_search

brave_spellcheck

brave_llm_context

brave_chat_completions

brave_local_place_search

brave_local_pois

brave_local_descriptions

brave_summarizer_search

brave_summarizer_summary

brave_summarizer_enrichments

brave_summarizer_followups

brave_summarizer_entity_info

brave_summarizer_title

`brave_web_search`

`brave_news_search`

`brave_image_search`

`brave_video_search`

`brave_suggest_search`

`brave_spellcheck`

`brave_llm_context`

`brave_chat_completions`

`brave_local_place_search`

`brave_local_pois`

`brave_local_descriptions`

`brave_summarizer_search`

`brave_summarizer_summary`

`brave_summarizer_enrichments`

`brave_summarizer_followups`

`brave_summarizer_entity_info`

`brave_summarizer_title`