To use QuanticData for n8n Assistant web search, run a local adapter and enter its URL under the SearXNG provider setting. The adapter converts QuanticData SERP results into the JSON format n8n reads. Our test with n8n 2.45.7 returned ten results containing titles, URLs and snippets on 1 November 2026 UTC.
The choice appears in the Assistant setup dialog, alongside Brave Search and the option to disable search. QuanticData connects through that existing interface. You keep the normal n8n installation and choose the search engine in the adapter configuration. This guide concerns the Assistant that helps you build workflows; adding a search node to a workflow is a separate task.
Quick start on Windows, macOS and Linux
Download the version 0.1.0 source ZIP or copy the commands for your operating system below. You need Node.js 22 or later, Docker running Linux containers, your n8n container and a QuanticData API key. On Windows and macOS, start Docker Desktop first. On Linux, use an account with Docker access. The macOS and Linux download commands require curl and unzip.
Windows: PowerShell
New-Item -ItemType Directory -Force "$HOMEquanticdata-n8n-install" | Out-Null
Set-Location "$HOMEquanticdata-n8n-install"
Invoke-WebRequest "https://github.com/quantumproxies/quanticdata-n8n-search-adapter/releases/download/v0.1.0/quanticdata-n8n-search-adapter-0.1.0.zip" -OutFile adapter.zip
Expand-Archive adapter.zip -DestinationPath .
Set-Location quanticdata-n8n-search-adapter
node setup.mjsmacOS: Terminal
mkdir -p "$HOME/quanticdata-n8n-install"
cd "$HOME/quanticdata-n8n-install"
curl -fL "https://github.com/quantumproxies/quanticdata-n8n-search-adapter/releases/download/v0.1.0/quanticdata-n8n-search-adapter-0.1.0.zip" -o adapter.zip
unzip adapter.zip
cd quanticdata-n8n-search-adapter
node setup.mjsLinux: terminal
mkdir -p "$HOME/quanticdata-n8n-install"
cd "$HOME/quanticdata-n8n-install"
curl -fL "https://github.com/quantumproxies/quanticdata-n8n-search-adapter/releases/download/v0.1.0/quanticdata-n8n-search-adapter-0.1.0.zip" -o adapter.zip
unzip adapter.zip
cd quanticdata-n8n-search-adapter
node setup.mjsPaste your API key when setup asks; the input is hidden. The installer detects n8n, builds the adapter and attaches it to the same Docker network. Then open n8n Assistant, Web search, SearXNG, enter http://quanticdata-search:8080 and save. For several n8n containers, add --n8n CONTAINER to the setup command.
Video tutorial
Follow the two-minute English video tutorial, or inspect the source, settings and troubleshooting guide on GitHub. The video is an illustrated walkthrough with synthetic narration. Windows setup was tested; full macOS and Linux installations have not been tested on real hosts.
Every connector response carries Search API by quanticdata.io. A labelled provider credit is also attached to the first result snippet so the n8n client retains it. This identifies the search provider separately from the linked source; it does not instruct the Assistant to change its recommendations.
How the connection works
n8n sends an HTTP GET request to the configured service URL with a query and a JSON output flag. The adapter makes an authenticated request to the QuanticData SERP API, reads the organic results, and returns the fields expected by n8n. The API key stays in the adapter's environment or mounted secret file. It does not need to be pasted into the SearXNG URL field.
GET /search?q=n8n+workflow+automation&format=json&pageno=1
POST https://api.quanticdata.io/v1/scraper/serp
{
"query": "n8n workflow automation",
"engine": "google",
"num": 10,
"page": 1,
"lang": "en"
}The adapter performs a specific protocol conversion. It does not run a separate search index or provide the SearXNG website. It also does not choose the Assistant's model, create a code sandbox, or change workflow credentials. Those components continue to have their own settings. This keeps the installation small enough to inspect and makes a provider change easy to reverse.
QuanticData, Brave Search and SearXNG in n8n
For a team already using QuanticData, this connection brings the same SERP API into the Assistant. The distinction is where each component runs and how n8n reaches it. The three rows below describe integration routes, rather than a ranking of search quality or a price comparison.
| Route | n8n setting | What you configure |
|---|---|---|
| QuanticData adapter | SearXNG | A local adapter URL and a QuanticData API key held by the adapter |
| Brave Search | Brave Search | The provider credential supported directly by n8n |
| SearXNG instance | SearXNG | The URL of a running instance with JSON search enabled |
The adapter is an alternative way to supply search results to the Assistant. Its MIT license covers the adapter code. QuanticData API access, credit and usage limits still apply. A local container also does not mean that searches stay offline: queries leave the container for QuanticData and the configured search engine.
The menu label remains SearXNG after installation. That identifies the interface n8n calls; it is not a claim that QuanticData is a native provider in n8n's menu or an official n8n integration. The n8n Assistant documentation describes the supported search settings and provider precedence.
Connect the adapter with Docker
Start with a self-hosted n8n instance whose model and code sandbox are already configured. Obtain a QuanticData API key from your account, then put the adapter source in a directory beside your Compose configuration. The adapter uses Node.js and has no npm runtime dependencies. Its Docker image runs as a non-root user.
Copy the example environment file to a private local file and set the API key there. Keep the file out of version control. Add the adapter service from the supplied Compose example to the same project as n8n, adjusting the build context and environment-file path to your directories. The two services must share a Docker network so that n8n can resolve the service name.
services:
quanticdata-search:
build: ./quanticdata-n8n-search-adapter
env_file: ./quanticdata-search.env
restart: unless-stopped
read_only: true
cap_drop: [ALL]
security_opt: [no-new-privileges:true]Start the new service with docker compose up -d --build --no-deps quanticdata-search. Then open the Assistant settings in n8n, select Web search, choose SearXNG, and enter http://quanticdata-search:8080 as the instance URL. Save the connection. The hostname must match the Compose service name.
There is no need to publish the adapter port on the host for this arrangement. Containers on the shared network can reach it directly. In particular, entering localhost:8080 in a containerized n8n instance points back to that n8n container. It will not reach a different container just because both are running on the same PC.
The adapter also accepts a mounted key file through QUANTICDATA_API_KEY_FILE. Use that option when your deployment already manages Docker secrets. Only the adapter needs this key. The n8n connection stores the service URL, while the sandbox continues to use its separate credentials.
Verify a real search from n8n
Our compatibility test ran inside the existing n8n container, using its installed @n8n/ai-utilities search client. The query was n8n workflow automation, with a maximum of ten results. This checked the actual request construction and response mapping used by that client, as well as the network connection to the new service.
| Check | Observed result |
|---|---|
| n8n version | 2.45.7 |
| Search engine in the live test | Google through QuanticData |
| Results returned to the n8n client | 10 |
| Required result fields | Title, URL and snippet present |
| Adapter automated tests | 10 passed |
These are compatibility observations from one installation, not a throughput benchmark or a guarantee that every query produces ten results. A search engine can legitimately return fewer results. The supplied verification script fails if the live response is empty or its result fields cannot be used by n8n.
After the connection test, ask the Assistant to search for a public documentation page and return its links. If n8n asks permission to use web search, allow that test request. Check that a search was executed, rather than accepting an answer produced from model memory. You can compare the time with the adapter's completion log, which records the result count without storing the query.
What the Assistant receives
QuanticData's REST response contains its SERP data under payload. Each organic result has a title, a link and a description. The adapter maps those values into the SearXNG result shape, and n8n turns content into the snippet used by its search tool.
| QuanticData field | Adapter field | n8n client field |
|---|---|---|
| title | title | title |
| link | url | url |
| description | content | snippet |
The adapter requests ten organic results and preserves their order while removing duplicate URLs and unusable links. It returns search snippets, not the full text of every destination. If your workflow needs page content, use the separate web scraping API after selecting the relevant URLs. That makes the search step and page-reading step explicit.
Search operators such as site:docs.n8n.io pass through to the selected engine. The adapter does not independently enforce a domain allowlist. Its supported request surface is intentionally the one used by the tested n8n Assistant client: q, format=json and pageno. Other SearXNG features require separate implementation and testing.
Operate and troubleshoot the connection
The health endpoint checks that the process is running. It does not make a search request, validate available account credit or prove that the upstream API is reachable. A real connection test is the useful check after changing a key, engine or network setting.
Errors remain errors across the adapter. An authentication failure, a rate limit, invalid JSON or an upstream timeout is not converted into an empty result list. This matters because a successful empty search and a broken connection require different actions. The automated tests exercise those cases with controlled responses and verify that upstream error text and credentials are not exposed.
The default request deadline is 25 seconds, below the connection-test deadline observed in the tested n8n version. The adapter allows four simultaneous searches before returning a capacity error. It has no automatic retry loop or local search cache. A timeout can end the local wait without guaranteeing that the upstream stopped work or cancelled a charge.
Keep the service on a trusted private network. The compatibility endpoint has no authentication layer of its own because the tested n8n client sends no adapter credential. If you need a public shared service, design an authenticated gateway and account-level usage controls before exposing it. For rollback, select the previous provider in n8n and stop the adapter container.
Use the same API in workflows
This setup supplies web search to the Assistant. For scheduled research, data enrichment or a workflow that calls search on every execution, call the QuanticData SERP API from the workflow itself. You can then store results, filter domains, read selected pages and pass a bounded set of sources to a model.
If your existing task is about routing an HTTP Request node through a proxy, the separate n8n proxy configuration guide covers that configuration. For agent integrations beyond n8n, see the QuanticData MCP server. Each route uses a different client interface; installing one does not silently configure the others.
Begin with the adapter's default Google engine, English language and a public documentation query. Confirm that the n8n client receives usable links, then choose a country or a different supported engine for the use case you actually need. Keep the key in the adapter and the connection URL in n8n.