Security context
What an agent needs to avoid regressing past fixes and find the next vuln in this repo.
mjpost/sacrebleu
master @ f615c72
2
Fixes
0
CVEs
MEDIUM
Peak severity
Highlights
Denial of Service: 1 prior fix. Scrutinize any change in this area.
sacrebleu/dataset/__init__.py: most-fixed (1 issue). Treat as high-risk during review.
0 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.
Insecure Transport: The dataset downloading mechanism historically pulled assets over cleartext HTTP, leaving the system vulnerable to transport-level tampering where an attacker could inject corrupted or malicious dataset payloads.
Denial of Service: The tokenization components utilized unbounded LRU caches, creating a vector for memory exhaustion attacks if subjected to high volumes of unique inputs. Capping the cache size prevents untrusted input from consuming all host memory.