Headless Protocol

Nuclei supports automation of a browser with simple DSL. Headless browser engine can be fully customized and user actions can be scripted allowing complete control over the browser. This allows for a variety of unique and custom workflows.

# Start the requests for the template right here
headless:

Actions

An action is a single piece of Task for the Nuclei Headless Engine. Each action manipulates the browser state in some way, and finally leads to the state that we are interested in capturing. Nuclei supports a variety of actions. A list of these Actions along with their arguments are given below:

navigate

Navigate visits a given URL. url field supports variables like {{BaseURL}}, {{Hostname}} to customize the request fully.

action: navigate
args: 
  url: "{{BaseURL}}

script

Script runs a JavaScript code on the current browser page. At the simplest level, you can just provide a code argument with the JS snippet you want to execute, and it will be run on the page.

action: script
args:
  code: () => alert(document.domain)

The code property strictly requires a function reference. Direct expressions or values are invalid and will not work. Always use a function.Incorrect:

action: script
args:
  code: alert(document.domain) # ❌ This is NOT a function reference

Correct:

action: script
args:
  code: () => alert(document.domain) # ✅ This is a function reference

Suppose you want to run a matcher on a JS object to inspect its value. This type of data extraction use cases are also supported with nuclei headless. As an example, let’s say the application sets an object called window.random-object with a value, and you want to match on that value.

- action: script
  args:
    code: () => window.random-object
  name: script-name
...
matchers:
  - type: word
    part: script-name
    words:
      - "some-value"

Nuclei supports running some custom Javascript, before the page load with the hook argument. This will always run the provided Javascript, before any of the pages load. The example provided hooks window.alert so that the alerts that are generated by the application do not stop the crawler.

- action: script
  args:
    code: () => (function() { window.alert=function(){} })()
    hook: true

This is one use case, there are many more use cases of function hooking such as DOM XSS Detection and Javascript-Injection based testing techniques. Further examples are provided on examples page.

click

Click simulates clicking with the Left-Mouse button on an element specified by a selector.

action: click
args: 
  by: xpath
  xpath: /html/body/div[1]/div[3]/form/div[2]/div[1]/div[1]/div/div[2]/input

Nuclei supports a variety of selector types, including but not limited to XPath, Regex, CSS, etc. For more information about selectors, see here.

rightclick

RightClick simulates clicking with the Right-Mouse button on an element specified by a selector.

action: rightclick
args: 
  by: xpath
  xpath: /html/body/div[1]/div[3]/form/div[2]/div[1]/div[1]/div/div[2]/input

text

Text simulates typing something into an input with Keyboard. Selectors can be used to specify the element to type in.

action: text
args: 
  by: xpath
  xpath: /html/body/div[1]/div[3]/form/div[2]/div[1]/div[1]/div/div[2]/input
  value: username

screenshot

Screenshots takes the screenshots of a page and writes it to disk. It supports both full page and normal screenshots.

action: screenshot
args: 
  to: /root/test/screenshot-web

If you require full page screenshot, it can be achieved with fullpage: true option in the args.

action: screenshot
args: 
  to: /root/test/screenshot-web
  fullpage: true

time

Time enters values into time inputs on pages in RFC3339 format.

action: time
args: 
  by: xpath
  xpath: /html/body/div[1]/div[3]/form/div[2]/div[1]/div[1]/div/div[2]/input
  value: 2006-01-02T15:04:05Z07:00

select

Select performs selection on an HTML Input by a selector.

action: select
args: 
  by: xpath
  xpath: /html/body/div[1]/div[3]/form/div[2]/div[1]/div[1]/div/div[2]/input
  selected: true
  value: option[value=two]
  selector: regex

files

Files handles a file upload input on the webpage.

action: files
args: 
  by: xpath
  xpath: /html/body/div[1]/div[3]/form/div[2]/div[1]/div[1]/div/div[2]/input
  value: /root/test/payload.txt

waitfcp

WaitFCP waits for the first piece of meaningful content, such as text or an image, indicating that the page is becoming useful.

action: waitfcp

waitfmp

WaitFMP waits for the First Meaningful Paint event, allowing users to proceed when content is visually ready.

action: waitfmp

waitdom

WaitDOM waits for the DOMContentLoaded event, indicating that the HTML has been loaded and parsed, but without waiting for stylesheets, images, and subframes to finish loading.

action: waitdom

waitload

WaitLoad waits the entire page, including dependent resources like stylesheets and images, has been fully loaded.

action: waitload

waitidle

WaitIdle waits until the page completely stopped making network requests and reaches a network idle state, indicating that all resources have been loaded.

action: waitidle

waitstable

WaitStable waits until the page is stable for N duration (default is 1s).

action: waitstable
args:
  duration: 5s

waitdialog

WaitDialog will wait for a JavaScript dialog (alert, confirm, prompt, or onbeforeunload) to be initialized and then automatically accept it.

action: waitdialog
name: alert
args:
  max-duration: 5s # (Optional. Default 10s.)

This action is useful for detecting triggered XSS payloads with a high level of accuracy and a low rate of false positives.

The name property MUST be explicitly defined to ensure the output variable is available for later use by matchers or extractors wihtin your template. See the example here.

Output variables:

NAME (boolean), indicator of JavaScript dialog triggered.
NAME_type (string), dialog type (alert, confirm, prompt, or onbeforeunload).
NAME_message (string), displayed message dialog.

getresource

GetResource returns the src attribute for an element.

action: getresource
name: extracted-value-src
args: 
  by: xpath
  xpath: /html/body/div[1]/div[3]/form/div[2]/div[1]/div[1]/div/div[2]/input

extract

Extract extracts either the Text for an HTML Node, or an attribute as specified by the user. The below code will extract the Text for the given XPath Selector Element, which can then also be matched upon by name extracted-value with matchers and extractors.

action: extract
name: extracted-value
args: 
  by: xpath
  xpath: /html/body/div[1]/div[3]/form/div[2]/div[1]/div[1]/div/div[2]/input

An attribute can also be extracted for an element. For example -

action: extract
name: extracted-value-href
args: 
  by: xpath
  xpath: /html/body/div[1]/div[3]/form/div[2]/div[1]/div[1]/div/div[2]/input
  target: attribute
  attribute: href

setmethod

SetMethod overrides the method for the request.

action: setmethod
args: 
  part: request
  method: DELETE

addheader

AddHeader adds a header to the requests / responses. This does not overwrite any pre-existing headers.

action: addheader
args: 
  part: response # can be request too
  key: Content-Security-Policy
  value: "default-src * 'unsafe-inline' 'unsafe-eval' data: blob:;"

setheader

SetHeader sets a header in the requests / responses.

action: setheader
args: 
  part: response # can be request too
  key: Content-Security-Policy
  value: "default-src * 'unsafe-inline' 'unsafe-eval' data: blob:;"

deleteheader

DeleteHeader deletes a header from requests / responses.

action: deleteheader
args: 
  part: response # can be request too
  key: Content-Security-Policy

setbody

SetBody sets the body for a request / response.

action: setbody
args: 
  part: response # can be request too
  body: '{"success":"ok"}'

waitevent

WaitEvent waits for an event to trigger on the page.

action: waitevent
args: 
  event: 'Page.loadEventFired'

The list of events supported are listed here.

keyboard

Keyboard simulates a single key-press on the keyboard.

action: keyboard
args: 
  keys: '\r' # this simulates pressing enter key on keyboard

keys argument accepts key-codes.

debug

Debug adds a delay of 5 seconds between each headless action and also shows a trace of all the headless events occurring in the browser.

Note: Only use this for debugging purposes, don’t use this in production templates.

action: debug

sleep

Sleeps makes the browser wait for a specified duration in seconds. This is also useful for debugging purposes.

action: sleep
args:
  duration: 5

Selectors

Selectors are how nuclei headless engine identifies what element to execute an action on. Nuclei supports getting selectors by including a variety of options -

Selector	Description
`r` / `regex`	Element matches CSS Selector and Text Matches Regex
`x` / `xpath`	Element matches XPath selector
`js`	Return elements from a JS function
`search`	Search for a query (can be text, XPATH, CSS)
`selector` (default)	Element matches CSS Selector

Matchers / Extractor Parts

Valid part values supported by Headless protocol for Matchers / Extractor are -

Value	Description
request	Headless Request
`<out_names>`	Action names with stored values
raw / body / data	Final DOM response from browser

Example Headless Templates

An example headless template to automatically login into DVWA is provided below -

id: dvwa-headless-automatic-login

info:
  name: DVWA Headless Automatic Login
  author: pdteam
  severity: high

headless:
  - steps:
      - args:
          url: "{{BaseURL}}/login.php"
        action: navigate
      - action: waitload
      - args:
          by: xpath
          xpath: /html/body/div/div[2]/form/fieldset/input
        action: click
      - action: waitload
      - args:
          by: xpath
          value: admin
          xpath: /html/body/div/div[2]/form/fieldset/input
        action: text
      - args:
          by: xpath
          xpath: /html/body/div/div[2]/form/fieldset/input[2]
        action: click
      - action: waitload
      - args:
          by: xpath
          value: password
          xpath: /html/body/div/div[2]/form/fieldset/input[2]
        action: text
      - args:
          by: xpath
          xpath: /html/body/div/div[2]/form/fieldset/p/input
        action: click
      - action: waitload
    matchers:
      - part: resp
        type: word
        words:
          - "You have logged in as"

More complete examples are provided here.

Templates

Protocols

Reference

Workflows

Actions

navigate

script

click

rightclick

text

screenshot

time

select

files

waitfcp

waitfmp

waitdom

waitload

waitidle

waitstable

waitdialog

getresource

extract

setmethod

addheader

setheader

deleteheader

setbody

waitevent

keyboard

debug

sleep

Selectors

Matchers / Extractor Parts

Example Headless Templates

​Actions

​navigate

​script

​click

​rightclick

​text

​screenshot

​time

​select

​files

​waitfcp

​waitfmp

​waitdom

​waitload

​waitidle

​waitstable

​waitdialog

​getresource

​extract

​setmethod

​addheader

​setheader

​deleteheader

​setbody

​waitevent

​keyboard

​debug

​sleep

​Selectors

​Matchers / Extractor Parts

​Example Headless Templates

Actions

navigate

script

click

rightclick

text

screenshot

time

select

files

waitfcp

waitfmp

waitdom

waitload

waitidle

waitstable

waitdialog

getresource

extract

setmethod

addheader

setheader

deleteheader

setbody

waitevent

keyboard

debug

sleep

Selectors

Matchers / Extractor Parts

Example Headless Templates