Back to skill

Security audit

测试专家评审

Security checks for vulnerabilities and agentic risk

Overview

The skill itself is a read-only QA review guide, but it recommends an unpinned bulk install of a third-party skill bundle.

Use the skill itself as a read-only QA review checklist if you are comfortable with Chinese instructions. Do not run the recommended bundle install unless you first inspect and trust the exact `Kokxi/qa-test-skills` revision and the installer it invokes.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (1)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:12
Finding
Unpinned Third-Party Skill Installation Creates a Supply-Chain Risk## Vulnerability Details **File Location**: `SKILL.md`, lines 12 and 54 **Vulnerability Type**: Unpinned third-party dependency installation **Risk Level**: Medium ### Vulnerable Code At line 12: ```text 本技能属于 QA Test Skills 技能集(49 个技能之一),完整工作流体验需安装全套:npx skills add Kokxi/qa-test-skills ``` At line 54: ```text > ⚠️ 本技能单独使用效果有限,建议配合完整技能集(12 步工作流)使用。安装:npx skills add Kokxi/qa-test-skills ``` Both instructions recommend executing: ```shell npx skills add Kokxi/qa-test-skills ``` ### Technical Analysis The installation command relies on mutable third-party components without specifying an immutable package version, repository commit, integrity digest, or signature. In particular: 1. `npx` may resolve and execute an externally maintained CLI package according to the local package-manager configuration. 2. `Kokxi/qa-test-skills` is referenced without a commit hash or other immutable revision. 3. The requested installation expands the trusted scope from the audited skill to a collection described as containing 49 skills. 4. The project provides no checksum, signature, lockfile, or documented verification procedure for that additional content. This does not establish that the current upstream content is malicious. However, it creates a supply-chain trust boundary in which future or compromised upstream content can differ from what was originally reviewed. ### Attack Path 1. An attacker compromises the relevant package, repository, maintainer account, publication process, or dependency used by the installer. 2. The attacker publishes a modified CLI release or changes the unpinned `Kokxi/qa-test-skills` content. 3. A user follows the installation recommendation in `SKILL.md`. 4. `npx` resolves the available installer, and the installer retrieves the mutable third-party skill bundle. 5. The compromised content is installed without verification against a reviewed immutable revision. 6. If the instal ...[truncated 1071 chars]
Remediation
## Remediation Suggestions 1. Pin the `npx` installer to an explicitly reviewed version rather than relying on implicit package resolution. 2. Pin `Kokxi/qa-test-skills` to an immutable commit hash or signed release tag. 3. Publish a cryptographic checksum or signature for the expected bundle and require verification before installation. 4. Replace the command with a documented two-stage process: download or inspect the exact pinned revision first, then install it only after review. 5. Audit all skills and scripts in the referenced bundle before recommending activation. 6. Run installation in a sandbox or least-privileged environment with restricted filesystem, network, credential, and tool access. 7. Document the exact additional files, scripts, hooks, and permissions introduced by the full bundle. 8. Use lockfiles or an equivalent dependency manifest to make transitive dependency resolution reproducible.
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (5)

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding
MCP server references in the skill manifest without version pinning are a rug-pull risk.

Natural-Language Policy Violations

Medium
Confidence
91% confidence
Finding
The display name and all user-facing instructions/examples are written exclusively in Chinese, with no indication that the user may choose another language. This can constitute a language/locale policy violation when the skill implicitly requires a specific language without documenting choice or justification.

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding
npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding
npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Vague Triggers

Medium
Confidence
95% confidence
Finding
The `when_to_use` field mixes a few specific phrases with broad natural-language conditions such as '需要对AI输出进行质量把关' and '用例上线前需要终审时'. These descriptions do not clearly bound when the skill should activate versus when other review or QA skills should be used, and they lack exclusion/negative examples.

Static analysis

No suspicious patterns detected.