cloudflare

Warn

Audited by Socket on Nov 3, 2026

2 alerts found:

SecurityAnomaly
SecurityMEDIUM
references/snippets/patterns.md

This fragment includes multiple request middleware patterns, most of which are conventional (security headers, geo redirect, A/B testing, bot blocking). However, the /api/ 'X-Internal-Auth' injection uses a hardcoded token and deletes the caller’s Authorization header before forwarding—an alarming authentication-manipulation behavior that could indicate sabotage/backdoor-style access control. Additionally, the maintenance bypass uses a hardcoded 'admin' token and CORS is permissive with wildcard origin and Authorization-related allowances. These elements warrant strong scrutiny before use.

Confidence: 78%Severity: 78%
AnomalyLOW
references/tail-workers/api.md

The provided code fragment is consistent with a telemetry forwarding tail handler and does not show overt malware or obfuscation. However, it unconditionally sends trace-derived request metadata—specifically `event.event?.request?.url`—to an external network endpoint via HTTP POST. If platform redaction is incomplete or `LOG_ENDPOINT` is not strictly trusted/controlled, this can constitute a meaningful sensitive-data exfiltration risk. Recommended mitigations: enforce destination allowlisting for `LOG_ENDPOINT`, omit or strongly sanitize/normalize the URL (e.g., drop query/identifiers), and rely on/redact-verify only the minimum fields required.

Confidence: 62%Severity: 56%
Audit Metadata
Analyzed At
Nov 3, 2026, 02:33 PM
Package URL
pkg:socket/skills-sh/openai%2Fplugins%2Fcloudflare%2F@1b5d31701dcc264c84ae124127bbd642fb2e199303089bdb66f525e58ab4c1a6