Case Studies

When the Chatbot Becomes the Vulnerability: AI Prompt Injection & Coupon Bypass in a Live E-Commerce VAPT

Full system prompt extraction, coupon generation restriction bypass (15% on 4 SKUs escalated to 100% discount on any product in the catalogue), and live database write access confirmed via AI function calling. All findings disclosed and remediated.

Critical

Redacted (E-Commerce, Mid-Market)

Techniques Used

100% Free

Coupon Bypass

2 weeks

SaaS / Productivity

Poisoned Knowledge: Indirect Prompt Injection via RAG in a B2B SaaS Support Assistant

Indirect prompt injection vector identified across the RAG ingestion pipeline before any customer was harmed. Untrusted user content removed from the indexable corpus and a content-trust pipeline introduced. System prompt extraction surface closed.

High

Redacted (B2B SaaS, ~40 employees)

RAG Poisoning

Attack Vector

< 2 days

Time to PoC

Marketing & Advertising

When the Agent Reads Its Own Tools: MCP Tool-Description Injection at a Marketing Agency

Tool-description injection vector identified before any client data left the environment. Third-party MCP servers pinned to vetted versions, sandboxed, and routed through a tool-description sanitiser. Internal allowlist of permitted tool calls introduced.

Critical

Redacted (Digital Marketing Agency, ~25 employees)

MCP Tool Description

Attack Vector

Any Prompt

Trigger

SaaS / Productivity

The Field That Wasn't on the Form: XHack AI Escalated a Standard Member to Workspace Owner via Mass Assignment

Mass-assignment privilege escalation surfaced autonomously by XHack AI and confirmed by human analysts. Role and tenant fields removed from the writable object; server-side field allow-listing applied to every update endpoint.

High

Redacted (Team Collaboration SaaS, ~150 employees)

< 48h

Time to Discovery

Workspace Owner

Privilege Gained

3 days

SaaS / MarTech

Hidden in Plain Sight: XHack AI Found a Live Email-Provider Key Baked Into a Survey Tool's JavaScript

Exposed transactional-email server token identified from the client-side bundle by XHack AI, confirmed live with a single read-only provider call, and rotated within hours. Email sending moved server-side; source maps removed from production.

High

Redacted (Customer-Feedback SaaS, ~10 employees)

JS Bundle

Entry Point

Email API Key

Secret Type

< 24 hours

Bug Bounty / VDP

SQLi & XSS in 100 Minutes: A Responsible Disclosure on Yogosha

Redacted (via Yogosha VDP)

5 vulnerabilities (3 SQLi, 2 Reflected XSS) identified and responsibly disclosed to the program owner within a single testing session. No exploitation beyond proof of concept.

~100 min

Time to Find

SQLi Found

XSS Found

1 hour 40 minutes

Retail

Zero-Trust Cloud Audit: Vulnerability Assessment for a Global Retailer

Multiple critical cloud misconfigurations identified across a 150k-user environment. PCI-DSS and GDPR compliance gaps closed before regulatory exposure.

150,000

Users at Risk

Critical

PCI-DSS

Compliance

Redacted (Global Retail)

4 weeks

E-Commerce / Retail

From 1,200 Daily Alerts to 31: How an E-Commerce Store Deployed XHack SOC and Caught a Credential Stuffing Campaign in Week One

Redacted (E-Commerce, D2C)

1,200+ daily alerts reduced to 31 actionable items. Active credential stuffing campaign detected and contained within 48 hours of SOC deployment. Mean time to detect dropped from unmeasured to 6 minutes.

97.4%

Alerts Reduced

6 min

Detection Time

Day 1

Active Threat Found

Ongoing (initial deployment: 3 hours)

FinTech / SaaS

One Number Off: Cross-Tenant IDOR in a Small Invoicing SaaS Exposed Every Customer's Invoices

Cross-tenant IDOR identified, scoped, and remediated within 48 hours. Tenant-aware authorisation introduced application-wide, not only on the affected route.

Critical

Redacted (B2B Invoicing SaaS, ~12 employees)

~12,000

Invoices Exposed

480

Tenants Affected

5 days

HR Technology / SaaS

From a Profile Picture to AWS Admin: SSRF via Image Fetch at an Early-Stage SaaS

SSRF chain to AWS IMDSv1 credential theft identified and stopped at proof-of-concept. Outbound URL fetch hardened, IMDSv2 enforced, IAM role tightened. No customer data left the environment.

Critical

Redacted (Early-Stage HR-Tech SaaS, ~20 employees)

Image Upload

Entry Point

AWS IAM Role

Credential Type

EdTech / SaaS

The Reset Link That Pointed at Us: Host-Header Account Takeover in a Mid-Market Learning Platform

Password-reset poisoning via Host and X-Forwarded-Host headers identified and demonstrated against XHack-controlled test accounts only. Reset links pinned to a fixed canonical URL; host headers no longer trusted. No real user was targeted.

High

Redacted (E-Learning / LMS, ~120 employees)

Password Reset

Attack Vector

Account Takeover

Impact

Marketing / Creative Services

A CV Upload to a Web Shell: Unauthenticated Remote Code Execution on a Small Agency's Careers Page

Unauthenticated file-upload leading to remote code execution identified and contained at a single benign proof file. Upload validation rebuilt around an allow-list, uploads moved off the web root, and execution disabled in the upload directory.

Critical