Security audit
Security checks for vulnerabilities and agentic risk
The skill is a shell-script helper, but its generated scripts can make high-impact system/deployment changes and unsafely embed user-provided values into executable Bash.
Review generated scripts carefully before running them, especially with sudo or deployment credentials. Do not pass untrusted or copied values into template parameters, and treat setup/deploy/monitor outputs as privileged operational scripts that may change servers, restart services, or delete/sync files.
scripts/shell.sh:25Command Injection in Generated Shell Scripts Through Unescaped Template Parameters
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
rsync -avz --delete \
--exclude='.git' \
--exclude='node_modules' \
--exclude='.env' \
--exclude='__pycache__' \
--exclude='.venv' \
./ "\${DEPLOY_USER}@\${DEPLOY_HOST}:\${DEPLOY_PATH}/" \
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
fi
# Check for common issues
grep -nE 'rm -rf /' "$file" && echo "🚨 危险: 检测到 rm -rf / ！" && ((issues++))
grep -nE 'eval ' "$file" && echo "⚠️ 使用了 eval，注意安全风险" && ((issues++))
# Check executable permission
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
fi
# Check for common issues
grep -nE 'rm -rf /' "$file" && echo "🚨 危险: 检测到 rm -rf / ！" && ((issues++))
grep -nE 'eval ' "$file" && echo "⚠️ 使用了 eval，注意安全风险" && ((issues++))
# Check executable permission
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
fi
# Check for common issues
grep -nE 'rm -rf /' "$file" && echo "🚨 危险: 检测到 rm -rf / ！" && ((issues++))
grep -nE 'eval ' "$file" && echo "⚠️ 使用了 eval，注意安全风险" && ((issues++))
# Check executable permission
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
fi
# Check for common issues
grep -nE 'rm -rf /' "$file" && echo "🚨 危险: 检测到 rm -rf / !" && ((issues++))
grep -nE 'eval ' "$file" && echo "⚠️ 使用了 eval,注意安全风险" && ((issues++))
# Check executable permission
YARA rule matched a known malware signature (reverse shell, backdoor, ransomware, C2 framework, or info stealer).
k unquoted variables
local unquoted
unquoted=$(grep -nE '\$[A-Za-z_]+[^"'"'"']' "$file" | grep -v '^#' | head -10 || true)
if [[ -n "$unquoted" ]]; then
echo "⚠️ 可能有未引号包裹的变量："
echo "$unquoted" | head -5 | sed 's/^/ /'
((issues++))
fi
# Check for common issues
grep -nE 'rm -rf /' "$file" && echo "🚨 危险: 检测到 rm -rf / ！" && ((issues++))
grep -nE 'eval ' "$file" && echo "⚠️ 使用了 eval，注意安全风险" && ((issues++))
# Check executable permission
if [[ ! -x "$file" ]]; then
echo "ℹ️ 文件没有执行权限，运行: chmod +x $file"
fi
# Try bash syntax check
if bash -n "$file" 2>/dev/null; then
echo "✅ 语法检查通过（bash -n）"
else
echo "❌ 语法错误："
bash -n "$file" 2>&1 | sed 's/^/ /'
((issues++))
fi
echo ""
if [[ $issues -eq 0 ]]; then
echo "✅ 未发现明显问题
The script presents itself as a developer workflow tool with commands like init, check, build, test, and deploy, but the implementation only prints reassuring status messages and logs invocations. In an agent or automation context, this can mislead users or higher-level systems into believing important validation, build, or deployment steps actually occurred, creating a security-relevant integrity gap and enabling skipped checks or unsafe releases.
The header frames the skill as a generator/checker, which implies producing or inspecting scripts rather than carrying out administrative changes itself. However, the generated setup template includes commands to upgrade packages, alter firewall rules, modify SSH configuration, and create swap, which is materially stronger operational behavior than the neutral 'checks' wording suggests.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
log "📊 当前备份数: \$TOTAL, 总大小: \$TOTAL_SIZE"
# ---- 可选：通知 ----
# curl -s -X POST "https://hooks.slack.com/services/YOUR/WEBHOOK" \\
# -H 'Content-type: application/json' \\
# -d "{\"text\": \"✅ 备份完成: \$BACKUP_FILE (\$SIZE)\"}"
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
\$RESTART_CMD
log "🔄 已执行重启命令"
# 通知（可选）
# curl -s -X POST "\$ALERT_WEBHOOK" -d "{\"text\": \"⚠️ \$APP_NAME 健康检查失败，已自动重启\"}"
else
log "✅ 健康检查通过"
fi
Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.
NC='\033[0m'
log() { echo -e "${GREEN}[设置]${NC} $*"; }
need_root() { [[ $EUID -eq 0 ]] || { echo "请用 sudo 运行"; exit 1; }; }
need_root
Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.
log "创建交换空间 (2GB)..."
if [[ ! -f /swapfile ]]; then
fallocate -l 2G /swapfile
chmod 600 /swapfile
mkswap /swapfile
swapon /swapfile
echo '/swapfile none swap sw 0 0' >> /etc/fstab
The setup template hard-codes timedatectl set-timezone Asia/Shanghai, which imposes a specific locale-related configuration on any server where the generated script is run. This is a natural-language/policy concern because the skill does not offer an alternative, prompt the user, or document that it is intended only for a China-specific deployment context.
The manifest description presents the skill in both Chinese and English, and the body repeats the same mixed-language presentation. This can violate a language/locale policy when the skill forces a specific language experience without explicitly letting the user choose their preferred language.
The comment labels the output as a 'service monitoring script', which suggests observation and logging. In practice, the script not only checks health but also executes a restart command automatically when the HTTP check fails, so the documentation understates that it is a remediation script as well as a monitor.
The markdown content is presented in Chinese ("Bash最佳实践") with no indication that the user can choose another language or that the skill is intentionally limited to a Chinese-speaking audience. This is a natural-language policy concern because it imposes a specific language without opt-in or justification.
No suspicious patterns detected.