Back to skill

Security audit

视频翻译配音 Video Translate & Dub

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed video subtitle and dubbing workflow that uses a third-party CLI and uploads the chosen media for processing, with some install and credential-storage caveats.

Install only if you are comfortable trusting the dLazy CLI and uploading the selected media to dLazy services. Prefer npx or an isolated project over a global install, use a dedicated working directory to avoid overwriting local outputs, and consider DLAZY_API_KEY for one-off use if you do not want a saved API key.

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:3
Finding
Unverified Third-Party CLI Installation and Execution<![CDATA[ ## Vulnerability Details **File Location**: `SKILL.md:3, 60-69` and mirrored instructions in `SKILL-cn.md:3, 60-69` **Vulnerability Type**: `T08: Insecure Dependencies` **Risk Level**: Medium ### Vulnerable Code ```yaml metadata: {"clawdbot":{"emoji":"🌐","requires":{"bins":["npm","npx","ffmpeg","ffprobe"]},"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"]}} ``` ```bash npx @dlazy/cli@1.2.3 <command> ``` ```bash npm install -g @dlazy/cli@1.2.3 ``` ### Technical Analysis The Skill directs the agent or user to download and execute the third-party npm package `@dlazy/cli`. Although the dependency is pinned to version `1.2.3`, the Skill does not verify a cryptographic integrity digest, vendor an audited copy, or otherwise bind execution to reviewed package contents. Version pinning prevents ordinary version drift but does not protect against compromise of the npm publisher account, registry infrastructure, or the already-published artifact. npm installation can also execute package lifecycle scripts. The global installation variant increases exposure by persistently adding package-controlled executables to the user's environment. No evidence in the audited files establishes that the referenced package is currently malicious. The finding concerns the unsafe trust and execution model required by the Skill. ### Attack Path 1. An attacker compromises the npm publisher account, package release, or another relevant supply-chain component. 2. The artifact delivered as `@dlazy/cli@1.2.3` is replaced or otherwise made to contain malicious package code or lifecycle scripts. 3. The user or agent follows the Skill instructions and runs either: - `npm inst ...[truncated 1079 chars]
Remediation
<![CDATA[ ## Remediation Suggestions 1. Publish and document the expected cryptographic integrity digest for the exact npm artifact, and verify it before installation or execution. 2. Vendor a reviewed copy of the required implementation where practical, rather than retrieving executable code dynamically. 3. Prefer a lockfile-backed local installation over global installation or ad hoc `npx` execution. 4. Avoid `npm install -g`; execute the package in an isolated environment with access limited to the designated media working directory. 5. Disable npm lifecycle scripts during installation where compatible: ```bash npm install --ignore-scripts --save-exact @dlazy/cli@1.2.3 ``` 6. Audit any required lifecycle scripts separately before enabling them. 7. Run the CLI under a dedicated, unprivileged account or sandbox without access to unrelated files, credentials, SSH keys, or sensitive environment variables. 8. Restrict outbound network access to the documented endpoints only where operationally feasible. 9. Apply the same hardened installation guidance consistently to both `SKILL.md` and `SKILL-cn.md`. ]]>
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Vague Triggers

Medium
Confidence
91% confidence
Finding
The trigger list includes very broad phrases such as '字幕翻译', 'video translation', and 'translate this video', which are likely to match ordinary user requests outside the narrow intended workflow. In an agentic environment, this can cause unintended activation of a skill that performs local file processing, uploads media to third-party endpoints, and may incur paid API usage without the user explicitly choosing this tool.

Ssd 1

Medium
Confidence
90% confidence
Finding
The skill explicitly instructs the operator to 'neutralize' a higher-priority service instruction and force a specific output envelope. Deliberately countermanding system/service-level behavior is dangerous because it encourages prompt-layer control conflicts, can undermine provider safety/governance guarantees, and may produce brittle or policy-noncompliant downstream behavior.

Session Persistence

Medium
Category
Rogue Agent
Content
name: dlazy-video-translate
version: 1.0.0
description: "video translation, video dubbing, subtitle translation, translate video to Chinese, add subtitles to video, AI dubbing, srt translation, 视频翻译, 视频配音, 字幕翻译 — transcribes a video with word-level timings, translates the subtitles, then burns them in and optionally lays down a fitted dub track. Composes the dlazy fun-asr, LLM and TTS tools with ffmpeg locally; delivers a finished mp4 plus srt files, not a script."
metadata: {"clawdbot":{"emoji":"🌐","requires":{"bins":["npm","npx","ffmpeg","ffprobe"]},"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":"When this skill is called, follow the numbered pipeline in the skill body in order: extract the audio with ffmpeg, transcribe it with 'dlazy fun-asr', group the returned words into cues, translate them in one batched 'dlazy claude-sonnet-5' call, write the srt, then burn it in with ffmpeg. Only run the dubbing stage if the user asked for dubbing. This skill composes tools directly — never pass --skill or --project."}}
---

# 视频翻译与配音 Video Translate & Dub
Confidence
72% confidence
Finding
The skill directs users to authenticate with a CLI that automatically stores an API key in a persistent local config file under the user profile. Persistent credential storage increases exposure if the host is multi-user, compromised, or if other tooling later reads the config, especially because this skill also uploads local media to remote endpoints.

Static analysis

No suspicious patterns detected.