Alibabacloud Tablestore Openclaw Memory

Security checks across static analysis, malware telemetry, and agentic risk

Overview

This is a coherent Alibaba Cloud memory setup, but it asks for broad cloud credentials and can change or bill cloud resources, so it needs review before use.

Install only if you are comfortable granting Alibaba Cloud access for persistent memory. Use a least-privilege RAM role or scoped temporary credentials, review the npm plugin before use, explicitly confirm any paid resource creation or public network exposure, and establish a way to delete or disable stored memories.

Static analysis

No static analysis findings were reported for this release.

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.

What this means

If these credentials are exposed or misused, they could give broad ability to administer and read/write Tablestore resources and use the DashScope account.

Why it was flagged

The setup asks the agent/user to handle long-lived cloud secrets and requires full-access Tablestore authority for a memory plugin, rather than clearly limiting credentials to one instance/table or a least-privilege policy.

Skill content
Credentials — AccessKey ID + Secret (with `AliyunOTSFullAccess` permission) ... [AGENT] Ask the user ... AccessKey ID ... AccessKey Secret ... 阿里云百炼 API Key
Recommendation

Prefer an ECS RAM role or short-lived, least-privilege RAM user scoped to the intended instance/table; avoid pasting long-lived secrets into chat; use environment variables or a secret store and declare the required credentials in metadata.

What this means

A Tablestore instance could become reachable over the public internet, increasing exposure even if authentication is still required.

Why it was flagged

Changing a cloud database instance to allow public internet access is a high-impact configuration change. The visible onboarding confirms new instance creation, but the supplied artifacts do not show a separate explicit approval for this network exposure fallback.

Skill content
UpdateInstance ... usage: "Enable public internet access when VPC endpoint is unreachable"
Recommendation

Require explicit user approval before enabling public internet access, prefer private/VPC endpoints, and document how to reverse the change.

What this means

The installed plugin code will run inside the user's OpenClaw environment and may receive the configured cloud credentials.

Why it was flagged

The skill installs an external npm package into OpenClaw. It is version-pinned and asks for consent, but the package contents, source repository, lockfile, and integrity hash are not included in the provided artifacts.

Skill content
openclaw plugins install @tablestore/openclaw-mem0@0.8.2 ... NPM_CONFIG_REGISTRY=https://registry.npmmirror.com
Recommendation

Verify the npm package publisher/source, prefer an integrity-pinned install, and review the plugin code before granting broad cloud credentials.

What this means

Information remembered by the agent may persist in Alibaba Cloud and be retrieved in later sessions.

Why it was flagged

The skill intentionally stores agent memory in an external cloud vector store for reuse across tasks. This is core to the purpose, but it affects privacy, retention, and future agent context.

Skill content
Tablestore Mem0 uses Alibaba Cloud Tablestore as the vector store backend for mem0, providing persistent long-term memory for AI agents.
Recommendation

Only store data appropriate for long-term cloud memory, define retention/deletion procedures, and confirm what user data is sent to Tablestore and DashScope.