Security context

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

nomic-ai/gpt4all

              main @ b666d16
            

Fixes

CVEs

HIGH

Peak severity

Highlights

Injection: 1 prior fix. Scrutinize any change in this area.

gpt4all-backend: 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.

Memory Corruption: Buffer overruns and heap out-of-bounds access in sampling functions (like repeat penalty sampling) can lead to arbitrary code execution or application crashes when processing model outputs with unchecked token windows.

Memory Safety: A use-after-free vulnerability can occur during embedding thread teardown if the model pointer is deleted out of sync with active background tasks, threatening application stability and state integrity.

Injection: Manual JSON construction via string concatenation allows malicious input to break the JSON structure, leading to injection attacks when sending data to the ingest API.