Security context

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

natebjones-projects/ob1

              main @ 671b923
            

Fixes

CVEs

HIGH

Peak severity

61.1%

Coverage

Highlights

Auth Bypass: 8 prior fixes. Scrutinize any change in this area.

integrations/enhanced-mcp/index.ts: most-fixed (3 issues). Treat as high-risk during review.

9 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: The use of SECURITY DEFINER on powerful database functions without revoking default public execution privileges allowed unauthenticated or untrusted users to query protected graphs directly via PostgREST. Ensuring functions default to SECURITY INVOKER or explicitly restrict execution to the service_role is necessary to prevent authorization bypass.

Prompt Injection: Processing untrusted email and user content without proper sandbox controls and prompt-hardening previously enabled escalation to command execution via the Codex CLI and model instructions manipulation. Implementing strict delimiters and removing dangerous execution flags is crucial when passing data to LLMs.

Information Disclosure: Passing sensitive API access keys as URL query parameters instead of in headers exposes credentials to logging infrastructures, proxies, and web history records. Transitioning all endpoints to header-based authorization is critical to preventing credential leakage.