portal-dev-setup

Warn

Audited by Socket on Sep 22, 2026

2 alerts found:

Securityx2
SecurityMEDIUM
SKILL.md

SUSPICIOUS. The workflow mostly aligns with a local portal-dev setup skill, but its actual footprint is high-risk: it requires opaque private Docker images from an unverifiable registry, reads raw credential locations, and includes a transitive eval-based dependency on another skill for npm auth. I do not see confirmed credential theft or covert exfiltration, so this is better classified as vulnerable/high-risk rather than outright malicious.

Confidence: 89%Severity: 82%
SecurityMEDIUM
scripts/onebox.sh

The script is primarily a legitimate Docker-based local development setup tool and contains no clear malware behavior such as exfiltration, persistence, or cryptomining. However, it has a serious command-injection vulnerability: attacker-controlled manifest fields can reach host-shell eval through cmd_add. Because manifests may be fetched from a configured remote catalog, a compromised catalog or malicious manifest could execute commands with the privileges of the user running the script. Do not use untrusted manifests until eval is removed and all identifiers and paths are strictly validated. The mutable latest tag and fixed development secret add operational security concerns but are explicitly documented development behavior.

Confidence: 98%Severity: 86%
Audit Metadata
Analyzed At
Sep 22, 2026, 04:21 AM
Package URL
pkg:socket/skills-sh/xgent-ai%2Fskills%2Fportal-dev-setup%2F@525d5101bad4500b35bccf0c595f34507be0a456696fc68c63ca98cef94736de
Security Audit — socket — portal-dev-setup