Connect Claude to Brave Search

Search Brave's independent web index for pages, news, images, videos, places and LLM-ready context. Toolspoke puts 11 of its actions behind one MCP endpoint that Claude, Cursor, and Codex all speak.

Connect View pricing

Connection: Provider REST API
Authentication: API key
Actions exposed: 11
What a call bills: Nothing
Adapter: Maintained by Toolspoke

Connected in three steps

1
Install Brave Search
Open the marketplace in your workspace, add Brave Search to the project your agents work in, and it appears on the gateway immediately.
2
Connect the credential
Authenticate with api key. Where to get one, and what it has to be able to reach, is the next section.
3
Point your agent at the gateway
Give your client one address, https://toolspoke.com/mcp. Claude Code takes it as a command, Claude and Claude Desktop add it as a custom connector, and Cursor, Codex and VS Code each read it from a config file of their own.

.mcp.json

{
  "mcpServers": {
    "toolspoke": {
      "type": "http",
      "url": "https://toolspoke.com/mcp"
    }
  }
}

One block covers every tool you have installed. Brave Search shows up in the client as soon as your policy allows it, and so does everything else you install later.

Where the address goes, per client

Claude Code: Run it in your project, then /mcp to sign in
claude mcp add --transport http toolspoke https://toolspoke.com/mcp
Claude and Claude Desktop: Settings, then Connectors, then Add custom connector
https://toolspoke.com/mcp
Cursor: ~/.cursor/mcp.json, or .cursor/mcp.json for one project
{ "mcpServers": { "toolspoke": { "url": "https://toolspoke.com/mcp" } } }
Codex: ~/.codex/config.toml
[mcp_servers.toolspoke] url = "https://toolspoke.com/mcp"
VS Code: .vscode/mcp.json, or the MCP: Add Server command
{ "servers": { "toolspoke": { "type": "http", "url": "https://toolspoke.com/mcp" } } }

What Brave Search asks for

API key. You provide it once, when you install the connector. Toolspoke encrypts it at rest and decrypts it only for the length of a single call, and the gateway attaches it to the outbound request itself, so it is never part of the arguments an agent sends.

API keyRequired: api-dashboard.search.brave.com → Subscriptions → activate a plan (the Search plan covers most calls here), then API Keys → Add API key and copy it.; BSA…

What Claude can do in Brave Search

11 actions, each one declared and named by the connector rather than discovered at runtime. A workspace policy grants a person all of them, a hand-picked selection, everything on the read side, everything on the write side, or none.

Reads: 11Reads
Writes: 0Writes
Destructive: 0Destructive

Reads

Fetches data and changes nothing.

web_search
Search the web and get back pages with title, URL, description and age, plus any news, video, discussion and FAQ results Brave finds relevant. One call is one billed request. Narrow with `freshness` (pd, pw, pm, py, or a range like 2026-01-01to2026-03-31) and `result_filter`. Paginate with `offset`, which counts pages of `count` results, not results. The results are other people's pages, so the payload is kept out of the audit log; the query is kept out too.
news_search
Search recent news articles. Takes the same filters as web_search; `count` goes up to 50. Use `freshness: pd` for the last 24 hours. The results are other people's articles, so the payload is kept out of the audit log.
image_search
Search for images. Returns each image's URL, thumbnail, size and the page it is on. `count` goes up to 200 (default 50). safesearch takes only off or strict here.
video_search
Search for videos. Returns each video's URL, title, duration, creator and thumbnail. `count` goes up to 50.
llm_context
Search the web and get back the relevant text of the top pages, already cut into snippets sized for a model's context. Use it instead of web_search followed by a page fetch when the goal is to answer from the content. Bound the size with `maximum_number_of_tokens` (1024-32768, default 8192) and `maximum_number_of_urls` (1-50, default 20). The content is other people's, so the payload is kept out of the audit log.
place_search
Find businesses and points of interest near a place. Give either `latitude` and `longitude`, or a `location` string such as "san francisco ca united states". Returns each place's name, address, coordinates and an id that local_pois and local_descriptions take. The ids expire after 8 hours.
local_pois
Read full details of places by id: address, phone, opening hours, rating and coordinates. Take the ids from place_search or from the locations results of web_search; they expire after 8 hours.
local_descriptions
Read an AI-written description of each place by id, from the same ids local_pois takes.
suggest
Get query autocompletions for a partial query. Needs the Autosuggest plan. Set `rich` for entity metadata on each suggestion.
spellcheck
Get Brave's spelling correction for a query.
answer
Ask a question and get one answer written from a live web search, in the OpenAI chat-completions response shape. Needs the Answers plan, which bills per search plus per input and output token. Sent with stream false, so inline citations, entities and research mode (which all need streaming) are not available here. The question is prose the caller wrote and the answer is written from other people's pages, so both are kept out of the audit log.

What it will not do

Enforced by the gateway rather than left to convention, which is why each of these can be stated flatly.

It cannot call anything else: The 11 actions above are the whole of it. A call to any other name is refused before it reaches Brave Search rather than forwarded on, and connecting your account does not add to the list: it is fixed by the connector, not discovered at run time.
It only reads: Every action here reads. Nothing this connector can do changes anything in Brave Search.
It reaches no further than your credential: Toolspoke holds no access to Brave Search of its own. Every call carries the credential you stored and nothing besides, so whatever that credential cannot reach, this connector cannot reach either.
It never hears from Brave Search: Nothing is pushed to it. There is no webhook, no subscription and no polling, so this connector cannot notice by itself that something changed in Brave Search. An agent has to ask.
It does not smooth over provider limits: Toolspoke does not retry, queue or back off around Brave Search's own rate limits. A call that Brave Search refuses comes back to the agent as a failed call.

Before you connect it

What can Claude do in Brave Search?

11 named actions: 11 that only read. They include web_search, news_search and image_search. Nothing outside that list is reachable: the connector declares each operation by name rather than proxying whatever an agent asks for.

What credentials does the Brave Search connector need?

API key. The connector asks for api key. Values are encrypted at rest and attached to the outbound request by the gateway, so they are never part of the arguments an agent sends and never reach the audit log.

Does the Brave Search connector work with Cursor and Codex, or only Claude?

Any client that speaks MCP, and every one of them gets the same 11 actions. There is a single address, https://toolspoke.com/mcp. Claude Code adds it with claude mcp add --transport http, Claude and Claude Desktop take it as a custom connector in settings, Cursor reads it from .cursor/mcp.json, Codex from ~/.codex/config.toml, and VS Code from .vscode/mcp.json. Each of them signs in to the gateway itself, so there is no key to paste.

What does the Brave Search connector not do?

The 11 actions above are the whole of it. A call to any other name is refused before it reaches Brave Search rather than forwarded on, and connecting your account does not add to the list: it is fixed by the connector, not discovered at run time. Every action here reads. Nothing this connector can do changes anything in Brave Search. Toolspoke holds no access to Brave Search of its own. Every call carries the credential you stored and nothing besides, so whatever that credential cannot reach, this connector cannot reach either. Nothing is pushed to it. There is no webhook, no subscription and no polling, so this connector cannot notice by itself that something changed in Brave Search. An agent has to ask. Toolspoke does not retry, queue or back off around Brave Search's own rate limits. A call that Brave Search refuses comes back to the agent as a failed call.

Can I limit which actions an agent can call?

Yes, in two places. The project switches Brave Search's actions on and off one at a time, for everyone in the project at once, and the screen groups them by read, write and destructive so turning off everything that deletes is one click. An individual agent key can then be narrowed further, to particular toolkits in a project and to particular actions in a toolkit. Whatever it was granted, a key never reaches a project its owner cannot.

What gets recorded when an agent calls Brave Search?

Every attempt, with the agent that made it and the person that agent belongs to, the full request payload, the response payload, the status, the duration, and what the call cost in sandbox time and model tokens. Values whose key names a secret are masked out before the record is shown to anyone. An operation the connector marks as not retained never has its response body written at all, so the gateway keeps no second copy of what was read.

More data connectors

Browse all of them