Back to skill

Security audit

huawei-cloud-modelarts-resource-pool-management

Security checks for vulnerabilities and agentic risk

Overview

This skill needs review because it can misstate costs, enable automatic payment without clearly explaining it, and execute crafted region input as code.

Wait for fixes to the pricing helper's input handling and total calculation before running it. Use restricted IAM access, verify costs independently, explicitly choose automatic payment, and approve each deletion only after reviewing affected workloads and backups. External installer contents and live cloud behavior remain unverified.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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
Findings (10)

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 106)May include surrounding context.

md
entials in `~/.hcloud/config.json`). See [references/cli-installation-guide.md](references/cli-installation-guide.md) for details.

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 279)May include surrounding context.

md
| [references/cli-installation-guide.md](references/cli-installation-guide.md) | CLI installation and authentication guide |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 277)May include surrounding context.

md
| [references/dataflow-diagram.md](references/dataflow-diagram.md) | Mermaid data flow diagram |

The billing workaround enables automatic payment without disclosing its financial effect.

High
Category
Output Handling
Confidence
97% confidence
Finding

The workaround says to always include auto.pay=true even though the stated validation error only concerns missing subscription-period parameters. This can turn a parameter repair into an automatically paid subscription without explicit approval of that billing choice; general pricing and write confirmation reduce exposure but do not disclose this additional financial effect.

Content

No source excerpt is available for this finding.

Destructive workarounds omit warnings about workload disruption and potential data loss.

High
Category
Output Handling
Confidence
98% confidence
Finding

The workarounds recommend deleting pools or deleting and recreating nodes without explaining workload interruption, possible loss of node-local data, or the fact that recreation does not restore prior state. General write-confirmation requirements reduce accidental execution but do not ensure informed approval when a reset, resize, or network-deletion request expands into destructive resource removal.

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · references/cli-command-examples.md (reported line 960)May include surrounding context.

Read/Write Operation Classification

Read Operations (No Confirmation Required)

text
ListPools, ShowPool, ShowPoolMonitor, ShowPoolStatistics, ShowPoolRuntimeMetrics,

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · references/dataflow-diagram.md (reported line 58)May include surrounding context.

Rendering diagram...

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · references/verification-method.md (reported line 38)May include surrounding context.

md
## Read vs Write Operations

- **Read operations** (list, show, query): Execute without confirmation
- **Write operations** (create, delete, update, patch, reboot, etc.): Require explicit user confirmation
- **Destructive operations** (delete pool, delete node pool, batch delete nodes): Show warning + require confirmation

Pricing output can report a zero total despite the documented fail-closed guarantee.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

When the response omits the top-level amount, validation accepts a positive sum of per-product amounts, but the output formatter independently defaults the missing total to 0 and displays a zero-cost total. This violates the fail-closed pricing guarantee and could mislead users into approving chargeable resource creation or expansion; required user confirmation and visible per-product prices reduce, but do not eliminate, the financial risk.

Content

No source excerpt is available for this finding.