Security context
What an agent needs to avoid regressing past fixes and find the next vuln in this repo.
Giskard-AI/giskard
main @ dfe93f5
39
Fixes
2
CVEs
HIGH
Peak severity
80.6%
Coverage
Highlights
Auth Bypass: 20 prior fixes. Scrutinize any change in this area.
backend/src/main/java/ai/giskard/config/WebSocketSecurityConfig.java: most-fixed (3 issues). Treat as high-risk during review.
28 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.
Auth Bypass: STOMP WebSocket handlers historically skipped token validation depending on licensing checks, handled null header accessors unsafely, or lacked JWT interceptors entirely, allowing unauthenticated connections.
Auth Bypass: Broken access control logic and improper entity evaluation (e.g., incorrect string/object equality checks, non-transactional lazy relation evaluations, and combined guest/owner authorization checks) routinely allowed unauthorized global write permissions.
Code Injection: Unsafe rendering of user-controlled conformity rules through Jinja2's default Template() constructor permitted arbitrary code execution and sandbox escapes (CVE-2026-40320).