ClawHub
Back to skill

Security audit

alibabacloud-loongcollector-ops

Security checks across malware telemetry and agentic risk

Overview

This skill is a focused Alibaba Cloud SLS operations helper with disclosed cloud access and strong approval gates for changes and deletion.

Install this only if you intend to let the agent operate your Alibaba Cloud SLS resources through aliyun sls. Review every normalized diff and exact resource name before approving writes, never paste AK/SK secrets into chat, and treat delete or cleanup requests as high-impact even though the skill requires extra confirmation.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Lp3

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding
The skill explicitly instructs the agent to run `aliyun configure list`, shell scripts under `scripts/`, and to consult local reference files, which are file-read and environment-adjacent capabilities. If the platform does not declare these permissions, the runtime may still expose them implicitly, creating an authorization gap where the skill can access local state beyond what the permission model signals to reviewers and users.

Description-Behavior Mismatch

Medium
Confidence
88% confidence
Finding
The skill is scoped as managing ingestion for an already-deployed LoongCollector, but it documents broader project/logstore lifecycle commands including create, update, and delete. That expands the agent’s operational blast radius beyond the stated purpose, increasing the chance that a prompt or workflow could trigger unauthorized infrastructure changes or destructive actions unrelated to collector operations.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The file lists destructive commands such as delete-project, delete-log-store, delete-machine-group, and delete-logtail-pipeline-config, but does not require an explicit confirmation step before execution. In an agent skill, merely marking commands with risk levels is insufficient because a mistaken or adversarial prompt could cause irreversible deletion of logging infrastructure and loss of observability.

VirusTotal

59/59 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.