🚨 If you run quantitative trading scripts or AI trading bots, check this Python package tonight
LiteLLM, a Python library commonly used in AI development, has run into trouble again. This is its second supply chain attack this year.
🔹 The affected version is 1.107.0 on PyPI; malicious code was hidden in the compiled installation package
🔹 It can steal SSH keys, AWS/Azure/GCP cloud credentials, passwords saved in browsers, and AI tool configurations
🔹 It uses a hidden file to trigger whenever a Python process starts—even if you haven’t imported this package
🔹 macOS, Linux, and Windows are all affected
🔹 Security teams including JFrog, Socket, and SafeDep have issued warnings, and PyPI has removed the affected version
This library has over 92 million downloads a month, so the potential impact is significant.
Why should crypto people care? Many people run trading bots, quantitative strategies, and on-chain monitoring scripts in Python environments. The same machine often also holds exchange API keys and server login keys, while browsers store all kinds of passwords. This kind of attack steals your “keys”—it doesn’t require you to click any phishing links.
The official advice is clear: pin your version to the clean 1.106.2 release, rotate all credentials, and thoroughly clean up infected files.
Three things I do myself:
1️⃣ Give exchange API keys trading permissions only, not withdrawal permissions, and bind them to an IP allowlist
2️⃣ Keep the machine used to run scripts separate from the one used to store wallets
3️⃣ Pin dependency versions; don’t blindly upgrade everything to the latest version with one click
Honestly, the biggest risk in crypto sometimes isn’t the chart—it’s what you’ve installed on your own computer.
Do you regularly check the dependencies used by your scripts, or do you install them and never look back? Let’s talk in the comments 👇
#供应链攻击 #API安全 #量化交易 #CryptoSecurity
LiteLLM, a Python library commonly used in AI development, has run into trouble again. This is its second supply chain attack this year.
🔹 The affected version is 1.107.0 on PyPI; malicious code was hidden in the compiled installation package
🔹 It can steal SSH keys, AWS/Azure/GCP cloud credentials, passwords saved in browsers, and AI tool configurations
🔹 It uses a hidden file to trigger whenever a Python process starts—even if you haven’t imported this package
🔹 macOS, Linux, and Windows are all affected
🔹 Security teams including JFrog, Socket, and SafeDep have issued warnings, and PyPI has removed the affected version
This library has over 92 million downloads a month, so the potential impact is significant.
Why should crypto people care? Many people run trading bots, quantitative strategies, and on-chain monitoring scripts in Python environments. The same machine often also holds exchange API keys and server login keys, while browsers store all kinds of passwords. This kind of attack steals your “keys”—it doesn’t require you to click any phishing links.
The official advice is clear: pin your version to the clean 1.106.2 release, rotate all credentials, and thoroughly clean up infected files.
Three things I do myself:
1️⃣ Give exchange API keys trading permissions only, not withdrawal permissions, and bind them to an IP allowlist
2️⃣ Keep the machine used to run scripts separate from the one used to store wallets
3️⃣ Pin dependency versions; don’t blindly upgrade everything to the latest version with one click
Honestly, the biggest risk in crypto sometimes isn’t the chart—it’s what you’ve installed on your own computer.
Do you regularly check the dependencies used by your scripts, or do you install them and never look back? Let’s talk in the comments 👇
#供应链攻击 #API安全 #量化交易 #CryptoSecurity