Software Security Audit Results

Nobody has claimed this yet.

Hi all, I submitted this github for a security audit and the automated test we used (sonarQube) came up with some findings. Anything medium or higher typically needs to be addressed before the software is allowed to be used. If anything could be fixed I would be appreciative, otherwise it will make things difficult with future versions, etc.

check_id	severity	path	message	License	Vulnerability	Source	Shortlink
trailofbits.python.pickles-in-pytorch.pickles-in-pytorch	HIGH	RF3/modelforge/src/modelhub/inference_engines/rf3.py	Functions reliant on pickle can result in arbitrary code execution. Consider loading from `state_dict`, using fickling, or switching to a safer serialization method like ONNX	AGPL-3.0 license	Insecure Deserialization	https://semgrep.dev/r/trailofbits.python.pickles-in-pytorch.pickles-in-pytorch	https://sg.run/NwQy
trailofbits.python.numpy-in-pytorch-modules.numpy-in-pytorch-modules	MEDIUM	RF3/modelforge/src/modelhub/loss/af3_losses.py	Usage of NumPy library inside PyTorch `SubunitSymmetryResolution` module was found. Avoid mixing these libraries for efficiency and proper ONNX loading	AGPL-3.0 license		https://semgrep.dev/r/trailofbits.python.numpy-in-pytorch-modules.numpy-in-pytorch-modules	https://sg.run/9vxr
python.lang.security.audit.eval-detected.eval-detected	MEDIUM	RF3/modelforge/src/modelhub/metrics/lddt.py	Detected the use of eval(). eval() can be dangerous if used to evaluate dynamic content. If this content can be input from outside the program, this may be a code injection vulnerability. Ensure evaluated content is not definable by external sources.	Semgrep Rules License v1.0. For more details, visit semgrep.dev/legal/rules-license	Code Injection	https://semgrep.dev/r/python.lang.security.audit.eval-detected.eval-detected	https://sg.run/ZvrD
python.lang.security.audit.eval-detected.eval-detected	MEDIUM	RF3/modelforge/src/modelhub/metrics/lddt.py	Detected the use of eval(). eval() can be dangerous if used to evaluate dynamic content. If this content can be input from outside the program, this may be a code injection vulnerability. Ensure evaluated content is not definable by external sources.	Semgrep Rules License v1.0. For more details, visit semgrep.dev/legal/rules-license	Code Injection	https://semgrep.dev/r/python.lang.security.audit.eval-detected.eval-detected	https://sg.run/ZvrD
trailofbits.python.numpy-in-pytorch-modules.numpy-in-pytorch-modules	MEDIUM	RF3/modelforge/src/modelhub/model/layers/af3_diffusion_transformer.py	Usage of NumPy library inside PyTorch `AttentionPairBiasDiffusionDeepspeed` module was found. Avoid mixing these libraries for efficiency and proper ONNX loading	AGPL-3.0 license		https://semgrep.dev/r/trailofbits.python.numpy-in-pytorch-modules.numpy-in-pytorch-modules	https://sg.run/9vxr
python.lang.security.audit.non-literal-import.non-literal-import	MEDIUM	RF3/modelforge/src/modelhub/resolvers.py	Untrusted user input in `importlib.import_module()` function allows an attacker to load arbitrary code. Avoid dynamic values in `importlib.import_module()` or use a whitelist to prevent running untrusted code.	Semgrep Rules License v1.0. For more details, visit semgrep.dev/legal/rules-license	Improper Authorization	https://semgrep.dev/r/python.lang.security.audit.non-literal-import.non-literal-import	https://sg.run/y6Jk
trailofbits.python.automatic-memory-pinning.automatic-memory-pinning	MEDIUM	RF3/modelforge/src/modelhub/utils/datasets.py	If possible, it is better to rely on automatic pinning in PyTorch to avoid undefined behavior and for efficiency	AGPL-3.0 license	Dangerous Method or Function	https://semgrep.dev/r/trailofbits.python.automatic-memory-pinning.automatic-memory-pinning	https://sg.run/jz5N
python.lang.security.deserialization.pickle.avoid-pickle	MEDIUM	RF3/modelforge/src/modelhub/utils/inference.py	Avoid using `pickle`, which is known to lead to code execution vulnerabilities. When unpickling, the serialized data could be manipulated to run arbitrary code. Instead, consider serializing the relevant data as JSON or a similar text-based serialization format.	Semgrep Rules License v1.0. For more details, visit semgrep.dev/legal/rules-license	Insecure Deserialization	https://semgrep.dev/r/python.lang.security.deserialization.pickle.avoid-pickle	https://sg.run/OPwB

Review the listed findings in RF3/modelforge/src/modelhub/inference_engines/rf3.py, loss/af3_losses.py, metrics/lddt.py, model/layers/af3_diffusion_transformer.py, resolvers.py, utils/datasets.py, and utils/inference.py, starting with the HIGH pickle finding and the eval/import findings. Use the SonarQube findings as the checklist; done means the medium-or-higher findings are addressed and the audit no longer reports them.

Written by the indexing model from the issue text.

Description

Contributor guide

First steps

Research direction

Assessment