Back to skill

Security audit

huawei-cloud-activity-submit

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because its submission workflow includes unrelated software installation and updates, persistent changes, unpinned downloads, and a script that scanners could only inspect partially.

Consider avoiding installation until the unrelated CLI and gallery installer are removed or independently reviewed and pinned, and the partially interpreted submit.mjs has been fully assessed. If you proceed, consider the current user's permissions, cloud credentials, terminal logs, and the possibility of external reporting; the scan does not establish credential exfiltration or malicious intent.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (17)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The source introduces a separate quality CLI and bootstrap script as prerequisites even though submitting activity URLs does not require them. This unnecessarily expands the workflow's trust boundary to installation and persistent executable changes; this source documents the dependency but does not itself contain the bootstrap implementation.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The prerequisite text promotes keeping an unrelated CLI up to date on each run, extending a submission task into a mutable software-update trust relationship. The actual update implementation is not present in this source, but the documented behavior can introduce unreviewed code into a credential-bearing environment.

Content

No source excerpt is available for this finding.

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 21)May include surrounding context.

md
- **纯 Node.js 脚本**:`submit.mjs`(Node ≥ 18,无外部依赖),不使用 KooCLI。

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 97)May include surrounding context.

md
4. **执行提交**:运行 `submit.mjs`(见下)。参数与步骤字段对应(**两步均需 `--project-id` 与 `--region`**;本 skill 固定使用生产环境接口)。

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
脚本位于本 skill 目录:`submit.mjs`(纯 Node,Node ≥ 18,无需额外依赖)。

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 177)May include surrounding context.

md
- **Node.js ≥ 18**:本 skill 的提交脚本 `submit.mjs` 使用原生 `fetch()`,需 Node 18 或以上版本。

The API-submission skill downloads and executes an unrelated quality CLI installer and creates a persistent PATH wrapper.

High
Category
Excessive Agency
Confidence
99% confidence
Finding

The script downloads a mutable bootstrap and executes it with the current user's privileges and inherited environment, then can write a persistent executable wrapper into a writable PATH directory. HTTPS, an OBS hostname check, and SHA-256 comparison offer some protection, but the hash comes from the same configurable metadata authority rather than an independently trusted pin; compromise of that authority can therefore yield arbitrary code execution.

Content

No source excerpt is available for this finding.

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

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.

Manifest references MCP server without version pin: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

The manifest references a gallery workflow whose installer and remote skill are not pinned to reviewed immutable versions. The scanner's MCP terminology is inaccurate, but the underlying supply-chain risk is real and duplicates the installation risk elsewhere in this source.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

Although this is a skill installer rather than an MCP server, the referenced publication workflow invokes an unpinned npm package and installs a skill from an unpinned remote repository. Compromise of either upstream could introduce executable code or adversarial instructions into an environment containing cloud credentials and project files; user consent to publication does not authenticate these dependencies.

Content

No source excerpt is available for this finding.

The workflow expands API submission into deletion of cloud resources.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

The manifest limits this skill to activity-submission API calls and optional gallery publishing. L102 additionally instructs the agent to offer and execute deletion of pay-as-you-go resources, including ECS instances, which is a destructive cloud-management operation rather than submission or a cleanup reminder. Requiring user confirmation mitigates accidental execution but does not resolve the scope mismatch.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

The installation command uses unversioned npx skills and a mutable Git repository, then automatically accepts installation with -y. This creates a supply-chain execution and instruction-injection path, despite the legitimate, explicitly optional gallery-publication purpose.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

This reference concerns the same unpinned installer used by the gallery workflow, not an MCP server or a separate execution path. Allowing users to decide whether to retry a failed installation does not protect against a malicious upstream version when installation succeeds.

Content

No source excerpt is available for this finding.

An unrelated quality CLI is introduced as a prerequisite.

Medium
Category
Excessive Agency
Confidence
95% confidence
Finding

An unrelated telemetry CLI is presented as a prerequisite, with bootstrap installation and ongoing upgrades, despite not being necessary for the declared API operation. This encourages unnecessary software execution and external reporting in a workflow that handles cloud credentials, without separate informed consent.

Content

No source excerpt is available for this finding.

The guide introduces an automatically reporting, self-updating telemetry CLI unrelated to activity submission.

Medium
Category
Excessive Agency
Confidence
95% confidence
Finding

The guide instructs installation of unrelated telemetry software, enables reporting by default, and describes automatic upgrades on each run. This creates an unnecessary data-disclosure and mutable-code dependency; the source does not specify telemetry contents, so actual credential exfiltration cannot be established.

Content

No source excerpt is available for this finding.

Overly Broad Trigger

Low
Category
Trigger Abuse
Confidence
70% confidence
Finding

Overly Broad Trigger: description clause 'publish-work-to-gallery)。Use when the user asks to 提交作品、提交闯关成果、自动提交活动成果、提交沙箱链接/公网地址, or when a generated sandbox URL needs to be submitted for the 华为云开放能力闯关挑战。本 skill' activates on 'to', which is too short and may match unintended inputs

Content

No source excerpt is available for this finding.

The environment check prints an access credential without warning about exposure in terminal output.

Low
Category
Output Handling
Confidence
94% confidence
Finding

The environment check prints the real access-key identifier into terminal output, which may be retained in agent transcripts or CI logs. An AK alone is not the secret key and normally cannot authenticate by itself, so this is limited information exposure rather than demonstrated account compromise.

Content

No source excerpt is available for this finding.