Alibabacloud Sas Openclaw Security

Security checks across malware telemetry and agentic risk

Overview

This skill is coherent for Alibaba Cloud/OpenClaw security operations, but it is powerful because it uses cloud credentials and can run commands or install security components on ECS instances.

Install this only if you need Alibaba Cloud/OpenClaw security automation. Use a dedicated least-privileged RAM identity, restrict ECS RunCommand to intended instances, verify CLI/plugin installation sources, explicitly approve every remote shell or guardrail install command, and keep generated security reports private.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Risk analysis

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

#
ASI03: Identity and Privilege Abuse
Medium
What this means

The skill can act with the Alibaba Cloud permissions granted to the configured CLI profile.

Why it was flagged

The skill relies on the user's Alibaba Cloud CLI credential profile, which carries cloud-account authority.

Skill content
If not yet configured, run `aliyun configure` and follow the prompts. Credentials are stored in `~/.aliyun/config.json`.
Recommendation

Use a dedicated RAM user or role, avoid primary-account keys, grant only the listed minimum actions, and restrict ECS RunCommand permissions to the intended instances.

#
ASI02: Tool Misuse and Exploitation
Medium
What this means

A mistaken or unauthorized command could change, stop, or expose data on remote ECS machines.

Why it was flagged

The ECS client can send arbitrary shell command content to specified ECS instances through Cloud Assistant.

Skill content
command_content: 命令内容(明文,自动进行 Base64 编码) ... "--type", command_type, "--command-content", command_b64 ... "--instance-id"
Recommendation

Treat this as remote administrator access: preview the full command and target instances, require explicit user approval, avoid unnecessary bulk execution, and keep audit logs.

#
ASI04: Agentic Supply Chain Vulnerabilities
Low
What this means

The local CLI/plugin environment may change and execute provider-supplied code outside the skill package.

Why it was flagged

The setup flow depends on a remote installer and automatically updated aliyun CLI plugins.

Skill content
run `curl -fsSL https://aliyuncli.alicdn.com/setup.sh | bash` ... run `aliyun configure set --auto-plugin-install true` ... run `aliyun plugin update`
Recommendation

Install the aliyun CLI from official documentation, verify the download source, and consider controlled versioning or administrative approval for plugin updates in production.

#
ASI05: Unexpected Code Execution
Medium
What this means

Installing the guardrail may execute provider-generated shell code on selected ECS instances.

Why it was flagged

The AISC API returns a shell installation command used for the security guardrail workflow.

Skill content
响应字段为 InstallKey,\n        但其实际含义是一条完整的 shell 安装命令(install command)
Recommendation

Review the generated install command where possible, run it only on intended hosts, and confirm rollback or uninstall steps before broad deployment.

#
ASI10: Rogue Agents
Low
What this means

Target ECS/OpenClaw environments will have an ongoing security component installed, which may affect future operations.

Why it was flagged

The guardrail is intended to keep running after installation, creating persistent behavior on target machines.

Skill content
Install the Alibaba Cloud security guardrail plugin to add continuous protection capabilities to OpenClaw instances.
Recommendation

Install it only when continuous protection is desired, track where it is deployed, and document how to disable or remove it.

#
ASI06: Memory and Context Poisoning
Low
What this means

Generated reports may reveal hostnames, IPs, vulnerabilities, alerts, and other sensitive security posture data.

Why it was flagged

The report generator stores raw instance, vulnerability, baseline, and alert data in local output files.

Skill content
"instances": instances,\n        "vulns": vulns,\n        "baseline": baseline,\n        "alerts": alerts
Recommendation

Store output reports in a protected directory, redact them before sharing, and delete old reports when no longer needed.