Security context

What an agent needs to avoid regressing past fixes and find the next vuln in this repo.

stanfordnlp/stanza

              main @ db60d1b
            

Fixes

CVEs

CRITICAL

Peak severity

0.0%

Coverage

Highlights

Deserialization: 4 prior fixes. Scrutinize any change in this area.

stanza/models: most-fixed (2 issues). Treat as high-risk during review.

7 high-severity fixes in this history; regressions here are high-impact.

Recurring patterns

The bug types that recur here, drawn from past fixes, not open vulnerabilities.

Code Execution: The project relied on default PyTorch model deserialization, allowing attackers to execute arbitrary code via maliciously crafted model checkpoints. This was addressed under CVE-2024-51750 by enforcing strict weight-only deserialization constraints.

Deserialization: Insecure standard pickle deserialization was utilized to restore model checkpoints in the common trainer module, creating a direct vector for remote code execution prior to migration to safer serialized formats.