Back to skill

Security audit

社交轮播图设计 Social Carousel

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed dLazy image-generation workflow, but users should understand it installs or runs a third-party CLI, stores an API key, and uploads prompts or media to dLazy services.

Install only if you are comfortable using dLazy's hosted service and npm-distributed CLI. Prefer the npx option if you do not want a persistent global install, review the linked CLI source/package, and avoid uploading sensitive local media or prompts unless you intend to send them to dLazy. Rotate or revoke the API key from the dLazy dashboard if you stop using 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:7
Finding
Unverified External npm CLI Is Installed and Executed with User Privileges<![CDATA[ ## Vulnerability Details **File Location**: `SKILL.md:7` and duplicated in `SKILL-cn.md:7` **Vulnerability Type**: Third-party supply-chain exposure through an externally retrieved executable dependency **Risk Level**: Medium ### Vulnerable Code ```yaml metadata: {"clawdbot":{"emoji":"🤖","requires":{"bins":["npm","npx"]},"install":"npm install -g @dlazy/cli@1.2.3","installAlternative":"npx @dlazy/cli@1.2.3","homepage":"https://github.com/dlazy-ai/cli","source":"https://github.com/dlazy-ai/cli","author":"dlazyai","license":"see-repo","npm":"https://www.npmjs.com/package/@dlazy/cli","configLocation":"~/.dlazy/config.json","apiEndpoints":["api.dlazy.com","files.dlazy.com"]},"openclaw":{"systemPrompt":"当你需要使用此技能时,请严格遵循此技能提供的指南进行规划和执行。你可以通过调用 dlazy CLI 的各类生成模型(如 dlazy seedream-4.5 等)来完成实际的图片渲染。注意:Windows PowerShell 中不允许使用 `&` 或 `&&` 进行命令串联或后台运行,请单独且同步地执行命令。"}} ``` The dependency is also invoked through the following documented mechanism in `SKILL.md:45-51` and `SKILL-cn.md:45-51`: ```bash npx @dlazy/cli@1.2.3 <command> ``` The execution instructions at `SKILL.md:208` and `SKILL-cn.md:208` require the agent to run the CLI: ```text After user confirmation, you must use the terminal to execute a command (such as `dlazy seedream-4.5 --prompt "..."`), with only one generation command executed at a time. ``` ### Technical Analysis The Skill contains no local implementation of the `dlazy` executable. Instead, it directs the runtime to retrieve and execute `@dlazy/cli@1.2.3` from the npm ecosystem, either globally or through `npx`. Although the version is pinned, the audited project does not contain a vendored copy, lockfile, package digest, signature verification policy, or other mechanism that would allow the executable artifact to be matched to reviewed source code. A global npm installation can also execute package lifecycle scripts and place binaries into the user's global npm environment. The `npx` alternative avoids a persistent global binary but ...[truncated 2345 chars]
Remediation
<![CDATA[ ## Remediation Suggestions 1. Vendor the exact CLI source or distributable into a reviewable, trusted build process rather than downloading executable code during Skill use. 2. If npm distribution remains necessary, maintain a lockfile or approved-artifact manifest containing the exact package version and integrity digest. Verify the downloaded artifact before execution. 3. Establish provenance verification using signed releases, npm provenance attestations, and a documented mapping between the reviewed source commit and published package artifact. 4. Avoid global installation. Run the dependency in an isolated container or restricted subprocess with a dedicated working directory, minimal filesystem permissions, a constrained environment, and network access limited to required endpoints. 5. Disable unnecessary npm lifecycle scripts where compatible with the package, and review all required lifecycle behavior before enabling it. 6. Provide credentials only to the generation subprocess and only for the duration of the request. Restrict permissions on `~/.dlazy/config.json`, and support short-lived, narrowly scoped API tokens where possible. 7. Require explicit user consent before installing the dependency, transmitting prompts, or uploading local media. 8. Correct the documentation so every reference identifies version `1.2.3`, or update all installation commands and metadata to the actual reviewed release. Add an automated consistency check to prevent future version drift. 9. Document a dependency-update review procedure covering source changes, package contents, lifecycle scripts, transitive dependencies, and artifact integrity before changing the pinned version. ]]>

SkillSpector

By NVIDIA
Vulnerability Patterns
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • 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 (2)

Description-Behavior Mismatch

Medium
Confidence
89% confidence
Finding
The skill materially expands from a stated planning/design workflow into direct terminal command execution against an external SaaS, including installation and use of a CLI. That increases the trust boundary, enables network/file handling through the toolchain, and can cause an agent to perform side-effectful actions the user may not have clearly expected from a design-planning skill.

Intent-Code Divergence

High
Confidence
95% confidence
Finding
The document states that image-generation prompt rendering details are out of scope, but later mandates drafting prompts and executing generation commands. This contradiction can bypass policy or reviewer expectations by presenting the skill as low-risk planning guidance while actually instructing an agent to take operational actions, including authenticated external API use.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.