Serpwise AI

Actions define what modifications the gateway applies when a rule's conditions match. Every SEO change is an action.

Actions are the "what" part of a rule. When a rule's conditions match, the gateway executes all configured actions against the HTML response before returning it to the visitor.

A single rule can have multiple actions — for example, setting a canonical tag and injecting JSON-LD structured data in a single rule.

Meta Tags

`set_meta_robots`

Sets or replaces the <meta name="robots"> tag in the page <head>.

When to use: Control crawler indexing behavior for specific pages or URL patterns.

Parameter	Value
`content`	e.g. `noindex`, `nofollow`, `noindex, nofollow`, `index, follow`

<!-- Result -->
<meta name="robots" content="noindex, nofollow">

Common values:

noindex — tell all crawlers not to index this page
noindex, nofollow — don't index and don't follow links
index, follow — explicitly allow indexing (overrides a previous noindex from origin)

`set_meta_title`

Sets or replaces the <title> element.

When to use: Override page titles without touching the CMS or codebase — useful for A/B testing titles or fixing titles across a template.

Parameter	Value
`content`	The title string

`set_meta_description`

Sets or replaces the <meta name="description"> tag.

When to use: Add or fix meta descriptions on pages where the CMS doesn't provide them.

Parameter	Value
`content`	The description string

`set_meta_tag`

Sets any arbitrary <meta> tag by name. If a tag with the same name attribute already exists, it's replaced.

When to use: Add custom meta tags not covered by the specific actions above.

Parameter	Value
`name`	The `name` attribute (e.g. `author`, `theme-color`)
`content`	The `content` attribute value

`remove_meta_tag`

Removes a <meta> tag by its name attribute.

When to use: Strip meta tags injected by your CMS that shouldn't appear on certain pages.

Parameter	Value
`name`	The `name` attribute to remove

Links

`set_canonical`

Sets or replaces the <link rel="canonical"> tag.

When to use: Fix canonical tags across filtered/sorted/paginated URLs, or consolidate duplicate content.

Parameter	Value
`href`	The canonical URL

<!-- Result -->
<link rel="canonical" href="https://www.example.com/products/blue-widget">

Problems it solves:

E-commerce sites with faceted navigation (color, size filters) generating thousands of near-duplicate URLs
CMS platforms that set incorrect canonicals on paginated pages
Syndicated content that needs to point back to the original

`add_hreflang`

Adds a <link rel="alternate" hreflang="..."> tag for international SEO.

When to use: Implement or fix hreflang across a localized site without modifying every page template.

Parameter	Value
`lang`	BCP 47 language tag (e.g. `en`, `de`, `en-US`, `x-default`)
`href`	The URL of the alternate page for this language

<!-- Result -->
<link rel="alternate" hreflang="en-US" href="https://www.example.com/en-us/page">

Create multiple add_hreflang actions within a single rule to build a complete hreflang set.

`remove_hreflang`

Removes all <link rel="alternate" hreflang="..."> tags, or a specific one by language.

When to use: Strip incorrect hreflang tags injected by your platform before adding corrected ones.

Parameter	Value
`lang`	Language tag to remove, or `*` to remove all

HTTP Headers

`set_response_header`

Adds or replaces an HTTP response header.

When to use: Add security headers, cache-control directives, or custom headers without server config changes.

Parameter	Value
`name`	Header name
`value`	Header value

Examples:

X-Robots-Tag: noindex          ← prevents indexing via header (useful for non-HTML)
Cache-Control: max-age=3600    ← set caching behavior for CDN/browser
X-Content-Type-Options: nosniff

`remove_response_header`

Removes an HTTP response header.

When to use: Strip headers that expose server information or conflict with your intended behavior.

Parameter	Value
`name`	Header name to remove

Examples: Remove X-Powered-By, Server, or platform-specific headers.

Template-based Schema.org JSON-LD injection. The gateway controls @context and @type automatically — you supply only the type name and properties. If a JSON-LD block with the same @type already exists on the page, it is replaced rather than duplicated.

When to use: Add structured data across many pages without writing raw JSON-LD. Use inject_json_ld when you need full control over the schema object; use auto_schema when you want the gateway to handle the boilerplate.

Parameter	Value
`schemaType`	Schema.org type name (e.g. `Article`, `Product`, `FAQPage`, `BreadcrumbList`)
`schema`	Props object — the gateway merges these with `@context` and `@type`; any user-supplied `@context` or `@type` keys are ignored

Template Variables

Any {variable} in a string value is resolved at request time. Variables work in nested objects and arrays.

Variable	Resolves to
`{page_title}`	Current `<title>` text (reflects any prior `set_meta_title` action in the same rule)
`{page_description}`	`<meta name="description">` content
`{page_image}`	`<meta property="og:image">` content
`{domain}`	Domain name (e.g. `www.example.com`)
`{scheme}`	Protocol (`https` or `http`)
`{path}`	URL path (e.g. `/products/blue-widget`)
`{full_url}`	Full request URL
`{query}`	Query string (without `?`)

You can also define custom domain variables that work alongside system variables. See Domain Variables for details.

Structural Builders

FAQPage and BreadcrumbList accept simplified input shapes that the gateway transforms into the correct schema structure.

FAQPage — provide a questions array with question/answer pairs. The gateway expands these into the required mainEntity array of Question/Answer types.

{
  "schemaType": "FAQPage",
  "schema": {
    "questions": [
      { "question": "What is SerpWise?", "answer": "The edge proxy for the Agentic Web — purpose-built for SEO and GEO." },
      { "question": "Is it no-code?", "answer": "Yes." }
    ]
  }
}

BreadcrumbList — provide an items array with name/url pairs. The gateway builds itemListElement with auto-incrementing position values.

{
  "schemaType": "BreadcrumbList",
  "schema": {
    "items": [
      { "name": "Home", "url": "https://www.example.com/" },
      { "name": "Products", "url": "https://www.example.com/products/" },
      { "name": "Blue Widget", "url": "{full_url}" }
    ]
  }
}

Examples

Article with template variables:

{
  "schemaType": "Article",
  "schema": {
    "headline": "{page_title}",
    "description": "{page_description}",
    "image": "{page_image}",
    "url": "{full_url}",
    "publisher": {
      "@type": "Organization",
      "name": "Acme Corp",
      "url": "https://{domain}"
    }
  }
}

FAQPage:

{
  "schemaType": "FAQPage",
  "schema": {
    "questions": [
      { "question": "How does SerpWise work?", "answer": "It sits between your CDN and origin." }
    ]
  }
}

BreadcrumbList:

{
  "schemaType": "BreadcrumbList",
  "schema": {
    "items": [
      { "name": "Home", "url": "https://{domain}/" },
      { "name": "Blog", "url": "https://{domain}/blog/" }
    ]
  }
}

Result:

<script type="application/ld+json">
{"@context":"https://schema.org","@type":"Article","headline":"My Article Title",...}
</script>

`inject_json_ld`

Injects a JSON-LD <script> block into the page <head>.

When to use: Add Schema.org structured data (Product, Article, BreadcrumbList, Organization, FAQ, etc.) to pages where your CMS doesn't support it.

Parameter	Value
`schema`	A valid JSON-LD object

{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "Blue Widget",
  "description": "A high-quality blue widget",
  "brand": {
    "@type": "Brand",
    "name": "Acme Corp"
  }
}

Multiple inject_json_ld actions can be used in the same rule — each generates a separate <script type="application/ld+json"> block.

`modify_json_ld`

Modifies an existing JSON-LD block already present in the page. Targets a specific @type and merges or overwrites specific properties.

When to use: Augment JSON-LD that your CMS already outputs — add missing properties without replacing the entire schema block.

Parameter	Value
`type`	The `@type` to target (e.g. `Product`, `Article`)
`properties`	Key-value pairs to merge into the matched schema

HTML Injection

`inject_html`

Injects an arbitrary HTML snippet at a specific position in the document.

When to use: Add scripts, tracking pixels, banners, or any custom HTML without a code deployment.

Parameter	Value
`position`	Where to inject: `head_start`, `head_end`, `body_start`, `body_end`
`html`	The HTML string to inject

Position	Where it goes
`head_start`	Immediately after `<head>`
`head_end`	Immediately before `</head>`
`body_start`	Immediately after `<body>`
`body_end`	Immediately before `</body>`

Example — inject a tracking pixel before </body>:

<img src="https://analytics.example.com/pixel.gif" width="1" height="1" alt="">

`remove_element`

Removes all HTML elements matching a CSS selector from the document.

When to use: Strip elements injected by your CMS or theme that shouldn't appear on certain pages — for example, removing social share buttons from noindex pages, or stripping duplicate schema blocks.

Parameter	Value
`selector`	A valid CSS selector

Examples:

.social-share-buttons    ← remove by class
#cookie-banner           ← remove by ID
script[src*="analytics"] ← remove analytics scripts on specific pages
meta[name="generator"]   ← remove CMS fingerprinting meta tags

Find & Replace

`find_and_replace`

Performs text substitution across the entire HTML response body.

When to use: Fix repeated text patterns across pages — update old brand names, fix broken URLs in content, replace deprecated phrases, or rewrite specific strings that your CMS outputs incorrectly.

Problems it solves:

A brand rename where old name appears across thousands of pages
An incorrect URL pattern embedded in CMS templates
Deprecated rel attributes or aria labels in template markup

Parameter	Value
`search`	The text or regex pattern to find
`replace`	The replacement string
`isRegex`	`true` to treat `search` as a regex pattern; `false` for literal text
`caseSensitive`	`true` or `false` (default: `true`)

Literal text example:

Find:    Acme Widget Co.
Replace: Acme Corporation

Regex example with capture group:

Find:    href="http://old-domain\.com(/[^"]*)"
Replace: href="https://www.new-domain.com$1"
Regex:   true

Find & replace operates on the raw HTML string, not the parsed DOM. Be precise with your patterns to avoid unintended substitutions. Always test with the URL Simulator first.

Conversion Tracking

`fire_conversion_event`

Fires a server-side conversion event to connected tracking platforms (Meta CAPI, TikTok Events API, Google Ads Enhanced Conversions).

When to use: Track purchases, leads, add-to-cart, and other conversion events server-side — bypassing ad blockers and Safari ITP.

Parameter	Value
`event_type`	`purchase`, `add_to_cart`, `lead`, `page_view`, `initiate_checkout`, `add_payment_info`, `search`, `view_content`, `complete_registration`, `contact`, `subscribe`, `custom`
`extraction`	Configuration for extracting event data (value, currency, order ID) from the HTML response

Events are sent to all active integrations configured for the domain. A first-party cookie is set on the visitor to enable cross-session attribution.

Robots.txt

`modify_robots_txt`

Modifies the robots.txt response served by your domain.

When to use: Add, update, or replace robots.txt directives without server access — useful when your CMS generates an incorrect robots.txt.

Parameter	Value
`mode`	`append` — adds to existing content; `replace` — replaces entirely; `generate` — creates from scratch if absent
`content`	The robots.txt content to apply

This action only fires when the request path is exactly /robots.txt and the response content type is text/plain.

Actions