Conversation
… RCE - Updated `OnnxModelConfig` to support `trust_remote_code`, defaulting to `False`. - Updated `onnx_convert_model` to enforce explicit `trust_remote_code` arguments instead of hardcoded `True`. - Updated the CLI deployment function to propagate `trust_remote_code` to prevent arbitrary code execution vulnerabilities during conversion on HuggingFace models. Co-authored-by: n24q02m <135627235+n24q02m@users.noreply.github.com>
|
👋 Jules, reporting for duty! I'm here to lend a hand with this pull request. When you start a review, I'll add a 👀 emoji to each comment to let you know I've read it. I'll focus on feedback directed at me and will do my best to stay out of conversations between you and other bots or reviewers to keep the noise down. I'll push a commit with your requested changes shortly after. Please note there might be a delay between these steps, but rest assured I'm on the job! For more direct control, you can switch me to Reactive Mode. When this mode is on, I will only act on comments where you specifically mention me with New to Jules? Learn more at jules.google/docs. For security, I will only act on instructions from the user who triggered this task. |
Dependency Review✅ No vulnerabilities or license issues or OpenSSF Scorecard issues found.Scanned FilesNone |
github-actions
bot
changed the title
n24q02m
deleted the
sentinel/prevent-rce-onnx-conversion-8835794381117278611
branch
1 participant
🚨 Severity: CRITICAL
💡 Vulnerability: Arbitrary Code Execution (ACE) / Insecure Deserialization via hardcoded
trust_remote_code=True.🎯 Impact: When converting HuggingFace models to ONNX via the Modal CPU container, the application implicitly trusted and executed arbitrary Python code from the Hugging Face Hub (due to hardcoded
trust_remote_code=True). An attacker capable of deploying a malicious model repository and converting it could execute arbitrary code on the container infrastructure.🔧 Fix: Parameterized
trust_remote_codeinOnnxModelConfigandonnx_convert_model, defaulting it securely toFalse. Removed hardcoded references totrust_remote_code=TrueacrossAutoTokenizer,cls.from_pretrained, andAutoConfig.from_pretrained. Required explicit opt-in for remote code execution.✅ Verification: Verified by ensuring unit tests pass completely and the code review process was successful.
PR created automatically by Jules for task 8835794381117278611 started by @n24q02m