agent-tools

Fail

Audited by Socket on Mar 9, 2026

1 alert found:

Malware
MalwareHIGH
SKILL.md

The skill aligns with its purpose of orchestrating many AI apps via a cloud CLI, but it relies on a dangerous download-execute pattern (curl ... | sh) to install a third-party CLI from an external URL, and it implies handling credentials for various integrations without explicit secure handling guidance. Data flows to external APIs (image/video generation, web search, Twitter) are plausible but underspecified. Overall, the skill is suspicious due to supply-chain risk and unclear credential/data-flow safeguards; treat as requiring strict sandboxing, verifiable sources, and explicit credential management before use in production.

Confidence: 98%Severity: 55%
Audit Metadata
Analyzed At
Mar 9, 2026, 12:30 PM
Package URL
pkg:socket/skills-sh/neversight%2Fskills_feed%2Fagent-tools%2F@b31647813d70ed711232cd280b2bebec66cc6424