Security audit
Security checks for vulnerabilities and agentic risk
The skill is mostly an advisory cloud-architecture guide, but it requests command-execution authority without clear need or guardrails.
Review before installing. The content itself looks like cloud-architecture guidance, but the exec permission should be removed or tightly documented with user-confirmed, read-only operational checks. Do not give it cloud credentials or let it run commands unless you understand the exact command and environment impact.
The skill uses a Chinese display name and Chinese-only descriptive text, which imposes a specific language/locale in the user-facing skill definition. There is no indication that users can opt into another language or choose their preferred locale.
The skill is presented as an advisory cloud architecture guide, yet the manifest grants an exec capability that enables arbitrary command execution. This expands the attack surface far beyond the documented purpose and could allow prompt-influenced or malicious inputs to trigger local system actions on the agent host.
Arbitrary command execution is not necessary for a skill whose stated purpose is architecture guidance, migration planning, cost optimization, and governance advice. Unjustified execution rights increase the likelihood of abuse, accidental destructive actions, credential exposure, or host compromise if the agent interprets user content or embedded instructions unsafely.
The documentation claims the skill provides framework guidance rather than executable scripts, but the manifest still authorizes command execution. This mismatch is dangerous because users and platform reviewers may underestimate the skill's real capabilities, creating a trust gap that can hide risky behavior behind benign documentation.
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.
user_data = base64encode(<<-EOF
#!/bin/bash
yum install -y nginx
systemctl enable nginx
systemctl start nginx
EOF
)
No suspicious patterns detected.