CVE-2026-84366

Plaintext S3 requests expose credentials and data; MITM can tamper responses.

• CVSS 7.4
• 319
• Information Leakage
• Remote

Scrapy is a high-level web crawling and scraping framework for Python. Prior to 2.17.0, in scrapy/core/downloader/handlers/s3.py, Scrapy's S3DownloadHandler converts an S3-scheme bucket and key request into a plaintext HTTP request to the corresponding S3 endpoint unless request.meta["is_secure"] is explicitly enabled, then signs and sends the plaintext request with configured AWS credentials. A network attacker who can observe traffic between Scrapy and S3 can read the bucket and key path, AWS Authorization header, X-Amz-Security-Token when temporary credentials are used, S3 object contents, and S3 response headers. An active man-in-the-middle attacker can also modify the plaintext S3 response body, status code, and headers before Scrapy processes them, causing scraped-data poisoning, poisoned exports, HTTP cache poisoning when caching is enabled, or influence over later crawl targets through forged redirects or attacker-controlled links. Users making S3-scheme requests with AWS credentials are affected. This issue is fixed in version 2.17.0.

CVE

CVE-2026-84366

CWE

319

CVSS base score

7.4

Published

2026-09-01

OWASP

A02 Cryptographic Failures

Orthogonal defect classification

Interface

Code defect classification

Missing Check

Category

Information Leakage

Subcategory

Information Disclosure

Accessibility scope

Remote

Impact

Information Disclosure

Affected component

Scrapy

Fixed by upgrading

Yes

Solution

Upgrade Scrapy to version 2.17.0 or newer.

import urllib.parse
import requests

class S3DownloadHandler:
    def __init__(self, aws_access_key, aws_secret_key):
        self.aws_access_key = aws_access_key
        self.aws_secret_key = aws_secret_key

    def download(self, request):
        # VULNERABLE: sends S3 request over plaintext HTTP
        parsed = urllib.parse.urlparse(request.url)
        bucket, key = parsed.netloc, parsed.path.lstrip('/')
        endpoint = f"http://{bucket}.s3.amazonaws.com/{key}"
        resp = requests.get(endpoint, auth=(self.aws_access_key, self.aws_secret_key))
        return resp.content

import urllib.parse
import requests
import botocore.auth
import botocore.awsrequest
import botocore.credentials

class S3DownloadHandler:
    def __init__(self, aws_access_key, aws_secret_key):
        self.creds = botocore.credentials.Credentials(aws_access_key, aws_secret_key)

    def download(self, request):
        # FIX: forces HTTPS and signs the request before sending
        parsed = urllib.parse.urlparse(request.url)
        bucket, key = parsed.netloc, parsed.path.lstrip('/')
        endpoint = f"https://{bucket}.s3.amazonaws.com/{key}"
        aws_req = botocore.awsrequest.AWSRequest(method="GET", url=endpoint)
        signer = botocore.auth.SigV4Auth(self.creds, "s3", "us-east-1")
        signer.add_auth(aws_req)
        headers = dict(aws_req.headers)
        resp = requests.get(endpoint, headers=headers)
        return resp.content

@misc{vaitp:cve202684366,
  title        = {{Plaintext S3 requests expose credentials and data; MITM can tamper responses.}},
  author       = {Bogaerts, Fr\'ed\'eric and Ivaki, Naghmeh and Fonseca, Jos\'e},
  year         = {2026},
  note         = {VAITP Python Vulnerability Dataset, entry CVE-2026-84366},
  howpublished = {\url{https://netpack.pt/vaitp/vulnerability/CVE-2026-84366/}}
}