Back to skill

Security audit

状态转换测试

Security checks for vulnerabilities and agentic risk

Overview

The skill itself is a read-only QA testing guide, but it prominently recommends an unpinned command that installs a broader external skill set.

Review this skill before installing the full recommended QA skill set. The single inspected skill is low-privilege and purpose-aligned, but the suggested unpinned `npx` command would fetch broader external content that was not audited here; pin versions or inspect that package before running it.

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:8
Finding
Unpinned Third-Party Skill Installation Creates Supply-Chain Risk## Vulnerability Details **File Location**: `SKILL.md`, lines 8 and 38 **Vulnerability Type**: Unpinned third-party dependency installation **Risk Level**: Medium **Complete vulnerable snippet:** ```text npx skills add Kokxi/qa-test-skills ``` The command appears in the Skill description and again in the installation recommendation. ### Technical Analysis The documentation recommends using `npx` to execute a package-resolved CLI and install Skill content from the external `Kokxi/qa-test-skills` repository. Neither the CLI package version nor the external repository commit is pinned, and no integrity hash or signature is specified. Consequently, the command may retrieve content that differs from what was available when this Skill was reviewed. The external CLI and repository are not included in the audited project, so their behavior cannot be verified from the available artifact. If either distribution channel is compromised, later users could install modified instructions, scripts, or other components. ### Attack Path 1. An attacker compromises the package resolved by `npx`, the referenced repository, or an associated maintainer account. 2. The attacker publishes a malicious CLI release or modifies the externally hosted Skill collection. 3. A user follows the recommendation and runs: ```text npx skills add Kokxi/qa-test-skills ``` 4. `npx` resolves and executes the mutable third-party CLI, which retrieves mutable external content. 5. The compromised content is installed into the user's Skill environment. 6. When the installed Skills are loaded or invoked, any malicious instructions or executable components they contain may operate with the permissions granted to the relevant agent or installation process. This attack path depends on the user manually executing the recommended command and on compromise or malicious modification of an external dependency. ### Impact Assessment The immediate audited ...[truncated 770 chars]
Remediation
## Remediation Suggestions 1. Pin the `skills` CLI to a specifically reviewed version rather than allowing `npx` to resolve the latest release: ```text npx skills@<audited-version> add ... ``` 2. Pin the external Skill collection to an immutable, reviewed commit instead of a mutable repository reference. 3. Verify downloaded artifacts using a cryptographic digest or trusted signature before installation. 4. Use a lockfile or equivalent reproducible dependency manifest where supported. 5. Prefer bundling reviewed dependencies with the project when licensing and distribution requirements permit. 6. Document the external trust boundary and advise users to inspect the resolved package and downloaded Skill content before execution. 7. Run installation with least privilege in a sandboxed environment, without credentials or access to sensitive workspaces. 8. Re-audit the complete downloaded Skill set and its scripts before granting it command execution, network access, or write permissions.
Vulnerability Patterns
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • 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
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
88% confidence
Finding
The display name, description, and usage guidance are entirely in Chinese, and the file does not indicate that language is optional or user-selectable. Under the stated policy, forcing a specific language or locale without opt-in is a natural-language policy concern unless the locale restriction is clearly documented and justified.

Rp1

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding
The skill instructs users to install or invoke `npx skills` without pinning a specific version, which can cause execution of whatever package version is current at install time. This creates a supply-chain and reproducibility risk: a compromised upstream package, a newly introduced malicious version, or unexpected breaking behavior could be pulled into the user's environment.

Vague Triggers

Medium
Confidence
94% confidence
Finding
The `when_to_use` field contains broad natural-language triggers such as generic references to state testing and state changes, which could match ordinary conversation and cause unintended skill invocation. In an agent environment, overbroad auto-selection can route user requests into the wrong workflow, increasing the chance of incorrect outputs, prompt-scope confusion, or unsafe follow-on tool usage in more capable skills.

Rp1

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding
This line repeats an unpinned `npx skills` installation command, exposing users to the same supply-chain risk as elsewhere in the file. Because the command appears in prominent guidance text, users may copy-paste it directly, making accidental execution of an unintended or malicious package version more likely.

Static analysis

No suspicious patterns detected.