Security context

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

tanstack/ai

              main @ 037e15a
            

Fixes

CVEs

HIGH

Peak severity

Highlights

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

@tanstack/ai-isolate-cloudflare: most-fixed (1 issue). Treat as high-risk during review.

2 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.

Remote Code Execution: Fetching model metadata from external services and writing them to dynamic .ts files executed via import statements allows attackers to execute arbitrary shell payloads if the upstream registry is compromised or spoofed. Utilizing static JSON instead of executable TypeScript prevents this execution path.

Code Injection: Evaluating dynamically constructed code within a Cloudflare isolate boundary exposes the application to direct code injection. Tool names must be strictly validated with regular expressions before being embedded in context creation logic.