cloud-security
Audited by Socket on Sep 19, 2026
7 alerts found:
Securityx5Anomalyx2This skill is internally consistent with its stated purpose, but that purpose is an offensive cloud security capability set for an AI agent. It does not show covert exfiltration or deceptive routing in the skill text itself, yet it meaningfully increases operational risk by combining credentialed cloud tooling, metadata credential-theft guidance, lateral movement, and exfiltration/evasion techniques. Classify as SUSPICIOUS/HIGH-RISK rather than malware.
The fragment is primarily a cloud-security training document with defensive scanners and configuration examples. It contains explicit offensive SSRF and AWS metadata credential-extraction code that can retrieve and print temporary IAM credentials, creating a significant security concern if executed against an authorized or unauthorized target. There is no evidence of embedded malware, persistence, sabotage, or covert exfiltration in the supplied fragment, but the extraction functionality is dual-use and potentially dangerous.
This is an explicit offensive cloud privilege-escalation playbook. It is not obfuscated malware or executable package code, but it contains highly actionable techniques for unauthorized IAM manipulation, credential theft, persistence, privileged workload deployment, and detection evasion. Use should be restricted to authorized security assessments; the material presents a high security risk if executed against systems without authorization.
This is a security-training document with legitimate defensive material, but it contains actionable dual-use and offensive instructions. In particular, metadata credential access, external exfiltration examples, command-injection payloads, invocation flooding, and public-package registration for dependency confusion should be restricted to authorized testing environments and rewritten with safer non-exfiltrating placeholders. The document itself does not constitute malware or execute these actions.
The material is an offensive Azure privilege-escalation and post-compromise guide. It contains actionable instructions for managed identity token access, privileged role assignment, remote command execution through VM extensions, and possible secret disclosure from Function App settings. No hidden malware or obfuscation is present, and no external exfiltration endpoint is shown, but the commands can cause unauthorized access or code execution when run against a permitted environment.
The fragment is a plainly written offensive cloud post-exploitation guide. It contains no executable package code or embedded malware, but it documents actionable techniques for privilege escalation, persistence, audit evasion, credential creation, and network pivoting. It should be treated as high-risk security content and not executed outside an explicitly authorized assessment.
The material is an explicit AWS penetration-testing lab guide, not package malware. It contains deliberately dangerous commands that can expose data, create administrator credentials, access instance metadata, assume roles, and delete resources if executed outside an isolated and authorized lab account. The main risks are operational misuse and incomplete credential cleanup, not covert malicious behavior.