Search Operators: Filter Results with Inline Commands

Search operators are special commands you embed directly inside the q parameter value. They act as surgical filters that tell Google precisely where to look — restricting results to a specific domain, a particular file format, pages whose titles contain a keyword, and more. Unlike the separate API parameters (exactTerms, siteSearch, fileType), search operators live inside the query string itself, which makes them ideal for dynamically constructed queries and power-user search experiences.

How search operators work

You append an operator to your query string in the format operator:value — no spaces between the colon and the value. You can mix operators freely with regular keywords and boolean logic in the same q value.

# Restrict to a single domain and require a keyword in the title
GET https://www.googleapis.com/customsearch/v1
  ?key=YOUR_API_KEY
  &cx=YOUR_SEARCH_ENGINE_ID
  &q=site:developer.mozilla.org+intitle:CSS+flexbox

Not all operators are officially documented as supported by the Custom Search JSON API in all configurations. site:, filetype:, intitle:, inurl:, and intext: have the broadest support. Results from related:, cache:, and link: may be inconsistent or return no results depending on your engine’s scope and the query. Test each operator with your specific engine configuration.

Operator reference

`site:` — Restrict to a domain or subdomain

Returns only results from the specified domain, subdomain, or URL prefix. This is the most commonly used search operator.

# All indexed pages from developer.mozilla.org
&q=site:developer.mozilla.org

# Only blog subdomain
&q=site:blog.example.com+kubernetes

# Results from a domain matching a path prefix
&q=site:docs.github.com/actions+workflows

# Combine with keyword search
&q=site:stackoverflow.com+python+async+await

Use site: inline when you need per-query domain scoping on top of an engine that already searches the full web. For permanent domain restrictions, configure sites in the Programmable Search Engine Control Panel and use the siteSearch API parameter instead — it is more reliable.

`filetype:` — Restrict to a file type

Returns only documents with the specified file extension. Useful for surfacing PDFs, spreadsheets, presentations, and other non-HTML documents.

# PDF documents about climate policy
&q=climate+policy+filetype:pdf

# PowerPoint presentations on machine learning
&q=machine+learning+filetype:pptx

# Excel spreadsheets with financial data
&q=annual+report+filetype:xlsx

# Plain text files (common for technical specs and changelogs)
&q=api+specification+filetype:txt

Commonly supported file types: pdf, doc, docx, xls, xlsx, ppt, pptx, txt, rtf, csv, xml, json

You can achieve the same effect using the dedicated fileType API parameter: &fileType=pdf. The API parameter is generally more reliable than the inline operator for programmatic use. Use the inline operator when you want to let users type it directly into a search box.

`intitle:` — Require a word in the page title

Returns only pages whose HTML <title> tag contains the specified word.

# Pages whose title contains "tutorial"
&q=python+intitle:tutorial

# Pages whose title contains "review" about noise-canceling headphones
&q=intitle:review+noise-canceling+headphones

# Pages with "guide" in the title about REST APIs
&q=intitle:guide+REST+API+design

allintitle: — All listed words must appear in the title (alternative form):

# Title must contain both "beginner" and "guide"
&q=allintitle:beginner+guide+to+kubernetes

`inurl:` — Require a word in the URL

Returns only pages whose URL contains the specified string. This is particularly effective for targeting blog posts, documentation paths, and structured URL schemes.

# URLs containing "/blog/"
&q=inurl:blog+product+announcement

# URLs containing "docs" — useful for documentation searches
&q=inurl:docs+authentication+oauth2

# Find pages with "download" in the URL
&q=software+inurl:download

# Combine with site: for precision
&q=site:github.com+inurl:releases+python+library

allinurl: — All listed words must appear in the URL:

&q=allinurl:api+reference+authentication

`intext:` — Require a word in the page body

Returns only pages whose body text contains the specified word. Because most searches already look at body text, intext: is most useful when you want to ensure a specific technical term actually appears in the content rather than just in metadata or links.

# Pages that explicitly mention "GDPR" in their body text
&q=privacy+policy+intext:GDPR

# Pages whose body discusses "rate limiting" specifically
&q=API+design+intext:"rate+limiting"

allintext: — All listed words must appear in the body text:

&q=allintext:docker+compose+healthcheck

Returns pages that Google considers topically similar to the specified URL. This is most effective with well-known, high-traffic domains.

# Pages similar to techcrunch.com
&q=related:techcrunch.com

# Pages similar to a specific article
&q=related:example.com/blog/intro-to-graphql

related: works best as a standalone operator without additional keywords. Results depend entirely on Google’s internal page similarity scores and may be limited or absent for less prominent sites.

`cache:` — Retrieve Google’s cached version info

Returns the Google-cached version of a URL. In the API context, this appears as search results linking to Google’s cache of the specified page rather than performing a keyword search.

# Information about the cached version of a page
&q=cache:example.com/article/123

cache: is primarily a browser/search UI concept. In the Custom Search JSON API, its behavior may vary — the returned result links to Google’s cached snapshot if one exists. Do not rely on this operator for production cache-fetching workflows.

`link:` — Find pages linking to a URL

Returns pages that contain at least one link to the specified URL. Note that Google has substantially reduced the scope of link: results in recent years, so results may be sparse.

# Pages that link to example.com
&q=link:example.com

# Pages that link to a specific article
&q=link:docs.example.com/api/authentication

Combining operators

You can stack multiple operators and keywords in a single q value. Each additional operator narrows the result set further.

# PDFs about machine learning on academic sites
&q=site:arxiv.org+filetype:pdf+"machine+learning"+intitle:survey

# Blog posts about React performance with "hooks" in the URL
&q=site:medium.com+inurl:blog+intitle:React+performance+hooks

# Find competitor pricing pages
&q=site:competitor.com+inurl:pricing+OR+inurl:plans

# GitHub repos with "authentication" in the URL and "OAuth2" in the title
&q=site:github.com+inurl:auth+intitle:OAuth2

Full operator reference table

Operator	Syntax	What it does	Example
`site:`	`site:domain.com`	Restricts results to a specific domain or subdomain	`site:docs.python.org asyncio`
`filetype:`	`filetype:ext`	Restricts results to a file type/extension	`report filetype:pdf`
`intitle:`	`intitle:word`	Requires one word in the page `<title>`	`intitle:tutorial python`
`allintitle:`	`allintitle:word1 word2`	Requires all listed words in the page title	`allintitle:getting started guide`
`inurl:`	`inurl:string`	Requires the string to appear in the URL	`inurl:docs authentication`
`allinurl:`	`allinurl:word1 word2`	Requires all listed words in the URL	`allinurl:api reference`
`intext:`	`intext:word`	Requires the word in the page body text	`intext:GDPR compliance`
`allintext:`	`allintext:word1 word2`	Requires all listed words in body text	`allintext:docker compose`
`related:`	`related:url`	Finds pages similar to the specified URL	`related:wikipedia.org`
`cache:`	`cache:url`	Returns Google’s cached version of a URL	`cache:example.com/page`
`link:`	`link:url`	Finds pages linking to the specified URL	`link:example.com`

Operators vs. dedicated API parameters

Some operators have direct API-parameter equivalents. When generating queries programmatically, prefer the dedicated parameter — it is more explicit and easier to validate.

Inline operator	Equivalent API parameter	Notes
`site:example.com`	`siteSearch=example.com`	API parameter supports `siteSearchFilter` (include/exclude mode)
`filetype:pdf`	`fileType=pdf`	API parameter is parsed separately from `q`
(none)	`exactTerms=phrase`	No inline equivalent for hard required-phrase enforcement
`-term` in `q`	`excludeTerms=term`	API parameter is cleaner for programmatic exclusion
`OR` in `q`	`orTerms=term1+term2`	API parameter separates the OR group from the main query

Get Started

Core Concepts

Search Features

Guides

Help

Search Operators: Filter Results with Inline Commands

How search operators work

Operator reference

`site:` — Restrict to a domain or subdomain

`filetype:` — Restrict to a file type

`intitle:` — Require a word in the page title

`inurl:` — Require a word in the URL

`intext:` — Require a word in the page body

`cache:` — Retrieve Google’s cached version info

`link:` — Find pages linking to a URL

Combining operators

Full operator reference table

Operators vs. dedicated API parameters

​How search operators work

​Operator reference

​site: — Restrict to a domain or subdomain

​filetype: — Restrict to a file type

​intitle: — Require a word in the page title

​inurl: — Require a word in the URL

​intext: — Require a word in the page body

​related: — Find similar pages

​cache: — Retrieve Google’s cached version info

​link: — Find pages linking to a URL

​Combining operators

​Full operator reference table

​Operators vs. dedicated API parameters

How search operators work

Operator reference

`site:` — Restrict to a domain or subdomain

`filetype:` — Restrict to a file type

`intitle:` — Require a word in the page title

`inurl:` — Require a word in the URL

`intext:` — Require a word in the page body

`related:` — Find similar pages

`cache:` — Retrieve Google’s cached version info

`link:` — Find pages linking to a URL

Combining operators

Full operator reference table

Operators vs. dedicated API parameters