CrossLinked is a LinkedIn enumeration tool that uses search engine scraping to collect valid employee names from a target organization. This technique provides accurate results without the use of API keys, credentials, or even accessing the site directly. Formats can then be applied in the command line arguments to turn these names into email addresses, domain accounts, and more.
For a full breakdown of the tool and example output, checkout:
https://m8r0wn.com/posts/2021/01/crosslinked.html
git clone https://github.com/m8r0wn/crosslinked
cd crosslinked
pip3 install -r requirements.txt
Examples
Results are written to a 'names.txt' file in the current directory unless specified in the command line arguments. See the Usage section for additional options.
python3 crosslinked.py -f '{first}.{last}@domain.com' company_name
python3 crosslinked.py -f 'domain\{f}{last}' -t 45 -j 1 company_name
Usage
positional arguments:
company_name Target company name
optional arguments:
-h, --help show this help message and exit
-t TIMEOUT Max timeout per search (Default=20, 0=None)
-j JITTER Jitter between requests (Default=0)
-v Show names and titles recovered after enumeration
Search arguments:
-H HEADER Add Header ('name1=value1;name2=value2;')
--search ENGINE Search Engine (Default='google,bing')
--safe Only parse names with company in title (Reduces false positives)
Output arguments:
-f NFORMAT Format names, ex: 'domain\{f}{last}', '{first}.{last}@domain.com'
-o OUTFILE Change name of output file (default=names.txt
Proxy arguments:
--proxy PROXY Proxy requests (IP:Port)
--proxy-file PROXY Load proxies from file for rotation
Proxy Support
The latest version of CrossLinked provides proxy support through the Taser library. Users can mask their traffic with a single proxy by adding --proxy 127.0.0.1:8080
to the command line arguments, or use --proxy-file proxies.txt
for rotating source addresses.
http/https
proxies can be added in IP:PORT
notation, while SOCKS requires a socks4://
or socks5://
prefix.