● Bugcrowd Tech-OPS Team Member
● Part time Hacker & Bug hunter
● Writer at Securityidiots.com
● Ex-Full time Penetration Tester
These Slides were originally developed and presented by
Jason Haddix at Defcon 23 on August 6th
● Director of Technical Ops at Bugcrowd
● Hacker & Bug hunter
● #1 on all-time leaderboard bugcrowd 2014
Source of the Slides
What this talk’s about...
And…LOTS of memes…. only some of them are funny
Step 1: Started with my bug hunting methodology
Step 2: Parsed some of the top bug hunters’ research (web/mobile only for now)
Step 3: Create kickass preso
Topics? BB philosophy shifts, discovery techniques, mapping
methodology, common attack parameters, useful fuzz strings, bypass or
filter evasion techniques, new/awesome tooling
Note: All information is from Jason Haddix’s own methodology and public resource. No
information from the Bugcrowd platform is obtained!
Differences from standard testing
● looking mostly for
● not competing with
● incentivized for count
● payment based on sniff
● looking for vulns that
aren’t as easy to find
● racing vs. time
● competitive vs. others
● incentivized to find
● payment based on
impact not number of
Find the road less traveled
^ means find the application (or parts of an
application) less tested to avoid duplicate.
1. *.acme.com scope is your friend
2. Find domains via Google (and others!)
a. Can be automated well via recon-ng
and other tools.
3. Confirm the subdomain to be in Scope
4. Port scan for obscure web servers or
services (on all domains)
5. Find acquisitions and the bounty
6. Functionality changes or re-designs
7. Mobile websites
8. New mobile app versions
Port scanning is not just for Netpen!
A full port scan of all your new found targets will usually
● separate webapps
● extraneous services
● Facebook had Jenkins Script console with no auth
● IIS.net had rdp open vulnerable to MS12_020
nmap -sS -A -PN -p- --script=http-title dontscanme.bro
^ syn scan, OS + service fingerprint, no ping, all ports,
Just what the hack is SecList?
SecLists is the security tester's companion. It's a collection of multiple types of lists used
during security assessments, collected in one place. List types include usernames,
passwords, URLs, sensitive data patterns, fuzzing payloads, web shells, and many more.
The goal is to enable a security tester to pull this repo onto a new testing box and have
access to every type of list that may be needed.
This project is maintained by Daniel Miessler and Jason Haddix.
Directory Bruteforce Workflow
After bruteforcing look for other status codes indicating you are denied or require auth then
append list there to test for misconfigured access control.
GET http://www.acme.com - 200
GET http://www.acme.com/backlog/ - 404
GET http://www.acme.com/controlpanel/ - 401 hmm.. ok
GET http://www.acme.com/controlpanel/[bruteforce here now]
Auth (better be quick)
Auth Related (more in logic and priv sections)
● Make sure they are in scope before submitting
● User/pass discrepancy flaw
● Registration page harvesting
● Login page harvesting
● Password reset page harvesting
● No account lockout
● Weak password policy
● Password not required for account updates
● Password reset tokens (no expiry or re-use)
Session (better be quick)
● Failure to invalidate old cookies
● No new cookies on login/logout/timeout
● Never ending cookie length
● Easily reversible cookie (base64 most often)
Customizable Themes & Profiles via CSS
Event or meeting names
Imported from a 3rd party (think Facebook integration)
JSON POST Values (check returning content type)
File Upload names
Uploaded files (swf, HTML, ++)
Custom Error pages
fake params - ?realparam=1&foo=bar’+alert(/XSS/)+’
Login and Forgot password forms
Core Idea: Does the page look like it might need to call on stored data?
There exist some SQLi polyglots, i.e;
SLEEP(1) /*‘ or SLEEP(1) or ‘“ or SLEEP(1) or “*/
Works in single quote context, works in double quote context, works in “straight into query”
context! (Mathias Karlsson)
You can also leverage the large database of
fuzzing lists from Seclists here:
SQL Injection Observations
Blind is predominant, Error based is highly unlikely.
SQLMap is king!
● Use -l to parse a Burp log file.
● Use Tamper Scripts for blacklists.
● SQLMapper Burp plugin works well to instrument SQLmap quickly.
Lots of injection in web services!
Common Parameters or Injection points
Item number values
sorting parameters (i.e order, sort, etc)
JSON and XML values
Cookie values (really?)
Custom headers (look for possible
integrations with CDN’s or WAF’s)
REST based Services
Burp Suite Extension
Burp allows you to use a range of addons/extensions
which can be added from BAPP Store, you download
and add manually or you can program your own script
and add to Burp.
There are many cool Burp Extensions you can add to
your collection to help you automate many manual
tasks and make your life easier.
- Reflected Parameters
SSRF (Server-Side Script Request Forgery)
Core Idea : Is there any external resource accessed by any
parameter which could be controlled by us.
Polyglot : www.yoursite.com/your_resource
Simply capture the IP from which your resource is accessed. There we start,
once we get the IP and we confirm that the resource is accessed by server-
side, we are up with our game for SSRF.
SSRF Tools - Testing & Exploitation
Burp Scanner, other few scanners in market….
As we know SSRF does not need automated fuzzing, because once we confirm a resource is accessible from the Server-Side
we can confirm SSRF/XFPA.
Once we have confirmed SSRF, we can move on to further exploitation which includes the following but not limited to:
1. Internal Server/Port Scan
2. Access to File System
3. SSRF via 306 Redirects
4. Exploitation via other known Protocols
XML External Entity Injection
Core Idea : Trial & Error, find any XML upload request or
any request which takes XML in input body.
Not very commonly we finds an application functionality which is
dealing with XML inputs. But if we do, we might get lucky to find an
Here’s how it works, if the XML is getting parsed by the application and
the External entities in the DTD (Document Type declaration) is
resolved then it may lead to XXE. You can also try converting a JSON
endpoint request to XML and try XML Injections.
XXE Tools - Testing & Exploitation
As the vulnerability is in its early stages we do not have any specific tool that totally concentrate on finding or exploiting
XXE, but as per automated scanning/finding we have Burp scanner, other updated automated vulnerability scanner which
are able to find XXE.
<?xml version="1.0" encoding="ISO-8859-1"?>
<!DOCTYPE foo [
<!ELEMENT foo ANY >
<!ENTITY xxe SYSTEM "file:///dev/random" >]><foo>&xxe;</foo>
Can be used to read system files + Other attacks SSRF is capable of.
Local file inclusion
Core Idea: Does it (or can it) interact with the server file system?
Liffy is new and cool here but you can also use Seclists: Common Parameters or Injection points
Malicious File Upload ++
File upload attacks are a whole presentation. Try this one to get a feel for bypass techniques:
● content type spoofing
● extension trickery
● File in the hole! presentaion - http://goo.gl/VCXPh6
Malicious File Upload ++
This is an important and common attack vector in this type of testing
A file upload functions need a lot of protections to be adequately secure.
● Upload unexpected file format to achieve code exec (swf, html, php, php3, aspx, ++) Web
● Execute XSS via same types of files. Images as well!
● Attack the parser to DoS the site or XSS via storing payloads in metadata or file header
● Bypass security zones and store malware on target site via file polyglots
Everyone knows CSRF but the TLDR
here is find sensitive functions and
attempt to CSRF.
Burps CSRF PoC is fast and easy for
Many sites will have CSRF protection, focus on CSRF bypass!
● Remove CSRF token from request
● Remove CSRF token parameter value
● Add bad control chars to CSRF parameter value
● Use a second identical CSRF param
● Change POST to GET
Check this out...
Debasish Mandal wrote a python tool to automate finding CSRF bypasses called
Step 1: Enable logging in Burp. Crawl a site with Burp completely executing all
Step 2: Create a template...
Or focus on pages without the token in Burp:
CSRF Common Critical functions
Add / Upload file Password change
Email change Transfer Money /
Delete File Profile edit
CSRF N/A functions
Logout CSRF Public Forms
Forms that don’t make any change
Often logic, priv, auth bugs are blurred.
Testing user priv:
Here is how it should be:
1. admin has power
2. user has few permissions
And we are looking for functions which are
only meant for the admin and are accessible
1. Find site functionality that is restricted to certain
2. Try accessing those functions with lesser/other
3. Try to directly browse to views with sensitive
information as a lesser priv user
Autorize Burp plugin is pretty neat here...
Common Functions or Views
Add user function
Delete user function
start project / campaign / etc function
change account info (pass, CC, etc) function
customer analytics view
payment processing view
any view with PII
Insecure direct object references
IDORs are common place in bounties, and hard
to catch with scanners.
Find any and all UIDs
● negative values
● Attempt to perform sensitive functions
substituting another UID
○ change password
○ forgot password
○ admin only functions
Common Functions , Views, or Files
Everything from the CSRF Table, trying cross account attacks
Sub: UIDs, user hashes, or emails
Images that are non-public
Private Files (pdfs, ++)
Shipping info & Purchase Orders
Sending / Deleting messages
Logic flaws that are tricky, mostly manual:
● substituting hashed parameters
● step manipulation
● use negatives in quantities
● authentication bypass
● application level DoS
● Timing attacks
A simple logic Flaw
An online cute dog contest, the dog with the best average of likes
1. Anyone can register and take part.
2. Once a dog is registered, people can start liking or disliking
3. Everyone dislikes each other’s dogs to win the contest
4. The dog with the best average wins the contest.
5. Registration and votings gets closed 5 minutes before the
results are announced.
What is the Logic Flaw over here?
The vulns formerly known as “noise”
● Content Spoofing or HTML injection
● Referer leakage
● security headers
● path disclosure
Things to take with you…
1. Crowdsourced testing is different enough to pay attention to
2. Crowdsourcing focuses on the 20% because the 80% goes quick
3. Data analysis can yield the most successfully attacked areas
4. A 15 minute web test, done right, could yield a majority of your critical vulns
5. Add polyglots to your toolbelt
6. Use SecLists to power your scanners
7. Remember to periodically refresh your game with the wisdom of other techniques and
Follow these ninjas who I profiled: https://twitter.com/Jhaddix/lists/bninjas
Tim Tomes - Recon-ng
Joe Giron - RFI params
Soroush Dalili - File in the Hole preso
Mathias Karlsson - polyglot research
Ashar Javed - polyglot/xss research
Ryan Dewhurst & Wpscan Team
Bitquark - for being a ninja, bsqli string
rotlogix - liffy LFI scanner
Arvind Doraiswamy - HTTPs, CSRF Burp Plugins
Barak Tawily - Autorize burp plugin
the RAFT list authors
Ferruh Mavituna - SVNDigger
Jaime Filson aka wick2o - GitDigger
Robert Hansen aka rsnake - polyglot / xss
Dan Crowley - polyglot research
Daniel Miessler - methodology, slide, and data contributions
My awesome team at Bugcrowd ( Ashley,Grant,Shpend,Fatih, Dan, Sean,Jay, Patrik ++)
Nullcon & All the bug hunting community!!!