Back to skill

Security audit

huawei-cloud-services-count

Security checks for vulnerabilities and agentic risk

Overview

The skill appears intended to count Huawei Cloud services, but it uses unsafe setup and credential instructions and auto-overwrites local KooCLI metadata for a simple count task.

Review before installing. Prefer installing KooCLI through a verified method, avoid placing AK/SK secrets directly on the command line, and run the count-only command against an existing metadata cache unless you explicitly want the agent to refresh and overwrite local metadata.

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

Description-Behavior Mismatch

Medium
Confidence
94% confidence
Finding
The documented flow performs `hcloud meta download`, which introduces network/state-changing behavior that contradicts the skill’s stated purpose of using offline cached metadata only. This can cause unexpected external communication, alter local cache state, and break user or platform assumptions about a read-only counting skill.

Intent-Code Divergence

Medium
Confidence
93% confidence
Finding
The flow description says the skill is for offline cached counting, but it also instructs a metadata download step, creating a misleading mismatch between claimed and actual behavior. Such documentation drift is security-relevant because reviewers and users may approve or invoke the skill under false assumptions about network access and side effects.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
Automatically piping 'y' into `hcloud meta download` suppresses an overwrite confirmation and causes a local state-changing action without an explicit user checkpoint. In an agent context, this increases the risk of unintended file modification, cache replacement, or destructive behavior being executed silently under the user's credentials.

Missing User Warnings

Medium
Confidence
98% confidence
Finding
The guide instructs users to pipe a remotely fetched script directly into bash, which executes unreviewed code from the network with no integrity verification or opportunity for inspection. If the hosting location, transport path, or upstream artifact is compromised, users could execute arbitrary code on their systems during installation.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The guide shows secrets being passed directly on the command line, which can expose credentials through shell history, process listings, terminal logging, or audit tooling. Compromise of the access key and secret key could enable unauthorized use of the user's Huawei Cloud account and associated resources.

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
- **Service**: Not applicable — `meta` is a KooCLI system command, not a service-specific command
- **Operation**: `download` — lower case (system operation)
- **Confirmation**: Pipe `echo "y" |` to auto-confirm the overwrite prompt
Confidence
93% confidence
Finding
The skill explicitly endorses autonomous confirmation of an overwrite prompt, allowing an agent to make a decision that changes local state without verifying user intent at execution time. This is dangerous in automation settings because it normalizes bypassing safety interlocks and can lead to silent modification of cached files or future extension to riskier commands.

Static analysis

No suspicious patterns detected.