Security context
What an agent needs to avoid regressing past fixes and find the next vuln in this repo.
lightly-ai/lightly-train
main @ 0706373
2
Fixes
0
CVEs
HIGH
Peak severity
Highlights
Deserialization: 1 prior fix. Scrutinize any change in this area.
and: most-fixed (1 issue). Treat as high-risk during review.
1 high-severity fix in this history; regressions here are high-impact.
Recurring patterns
The bug types that recur here, drawn from past fixes, not open vulnerabilities.
Deserialization: The use of unsafe deserialization during model loading allows arbitrary pickle execution. Untrusted checkpoints can exploit `torch.load` to achieve remote code execution (RCE) unless restricted to safe weights-only loading.
Man-in-the-Middle: Downloading model checkpoints without integrity verification exposes the client to Man-in-the-Middle (MitM) attacks. Rogue checkpoints could be swapped in transit, which subsequently triggers downstream deserialization issues.