opencode-openai-codex-auth
Warn
Audited by Socket on Sep 16, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS. The skill's core function is coherent with its stated purpose, and data appears intended for OpenAI endpoints plus local OpenCode config/token storage. The main risk is trust: a third-party personal publisher distributes an auth plugin that executes install-time package code and handles ChatGPT session tokens via an unpinned `npx @latest` flow. That is disproportionate enough to warrant caution, but there is not enough evidence here of credential exfiltration or confirmed malicious behavior.
Confidence: 88%Severity: 58%
Audit Metadata