抖音直播弹幕AI智能回复助手

• Outcome: Review
• Latest audit: May 29, 2026 at 9:26 PM UTC
• Version:
  0.1.0

Overview

The skill mostly matches its Douyin live-chat assistant purpose, but it runs a large obfuscated signing script, disables WebSocket certificate checks, and stores/forwards viewer chat data with limited privacy disclosure.

Install only if you are comfortable running an obfuscated Douyin signing script under Node. Use a virtual environment, provide the DeepSeek key via an environment variable, review or sandbox sign.js, restore certificate verification if possible, and treat saved chat logs/cache as viewer data that may require consent, protection, and periodic deletion.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Behavioral AST: exec() Call, eval() Call, Dynamic Import
• Taint Tracking: Direct Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands

Findings (3160)

subprocess module call

Medium

• Category: Dangerous Code Execution

Content

# 使用 Node.js 子进程执行 get_sign_wrapper.js，通过 stdin/stdout 传参
        wrapper_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'get_sign_wrapper.js')
        result = subprocess.run(
            ['node', wrapper_path],
            input=s,
            capture_output=True,

• Confidence: 70% confidence
• Finding: result = subprocess.run( ['node', wrapper_path], input=s, capture_output=True, text=True, encoding='utf-8', timeout=10

Tainted flow: 's' from requests.get (line 178, network input) → subprocess.run (code execution)

Critical

• Category: Data Flow

Content

# 使用 Node.js 子进程执行 get_sign_wrapper.js，通过 stdin/stdout 传参
        wrapper_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'get_sign_wrapper.js')
        result = subprocess.run(
            ['node', wrapper_path],
            input=s,
            capture_output=True,

• Confidence: 90% confidence
• Finding: result = subprocess.run( ['node', wrapper_path], input=s, capture_output=True, text=True, encoding='utf-8', timeout=10

Intent-Code Divergence

Low

• Confidence: 98% confidence
• Finding: At L5, the docstring states the function performs 'MD5 encryption'. However, the implementation at L6 uses hashlib.md5(...).hexdigest(), which is a one-way hash, not encryption. This is an intent/documentation mismatch that could mislead callers about the function's security properties and behavior.

Intent-Code Divergence

Medium

• Confidence: 97% confidence
• Finding: The docstring for `parseChatMsg` says `调用DeepSeek AI生成回复`, which implies the skill will actually reply to chat messages. In code, it generates a reply object, writes it to `REPLY_FILE`, and prints it locally, but never sends any outbound message through the websocket or another API. This is an intent-level contradiction between the documented behavior and the actual implementation.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: This markdown file states that the skill generates `ai_replies.jsonl` containing timestamps, usernames, danmu content, and replies, which is user data retention. Under the markdown-specific warning criteria, the description should clearly warn users about privacy and data-handling implications, but this section presents the files as routine output without any caution or disclosure.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The introduction says the skill captures Douyin live-room danmu in real time and uses DeepSeek AI to analyze user intent and generate replies, which implies third-party processing of user-generated content. The README explains functionality and setup, but does not clearly warn that live comments may be transmitted to an external API or discuss privacy implications.

Natural-Language Policy Violations

Low

• Confidence: 77% confidence
• Finding: Natural-language policy review applies to all file types, including markdown. This README presents the skill exclusively in Chinese with no indication that language is selectable or intentionally limited to a specific locale, which can be read as forcing a language/locale without user opt-in.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The markdown states that the system collects live barrage messages and uses the DeepSeek API to deeply analyze each message, which implies viewer content is sent off-platform to a third-party service. While the file mentions API keys and costs, it does not clearly warn users about the privacy impact of transmitting audience messages to an external provider.

Natural-Language Policy Violations

Low

• Confidence: 88% confidence
• Finding: The natural-language instructions, headings, and examples are all presented in Chinese, and the document does not offer an alternative language or indicate that the skill is intentionally limited to Chinese-speaking users. Under the stated policy, forcing a specific language without user opt-in can be a locale-policy issue.

Natural-Language Policy Violations

Medium

• Confidence: 92% confidence
• Finding: This markdown file contains natural-language guidance that assumes Chinese as the required interaction language across all example prompts. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale constraint is clearly documented and justified, which is not present here.

Natural-Language Policy Violations

Low

• Confidence: 89% confidence
• Finding: The docstring is written only in Chinese, which imposes a specific language choice in the skill's natural-language interface without offering an alternative or documenting a justified locale constraint. The policy for this review flags language or locale constraints that are forced without user opt-in.

Natural-Language Policy Violations

Medium

• Confidence: 86% confidence
• Finding: This code file contains user-facing natural-language instructions and comments entirely in Chinese, including setup guidance and configuration descriptions. Under the stated policy, forcing a specific language without offering a language or locale choice is a natural-language policy concern.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: This code sends the viewer username and chat message to the DeepSeek API, which is a network transmission of user data. Although the module docstring says it calls the DeepSeek API, there is no user-facing warning, confirmation, or privacy disclosure around sending chat content off-box, and the surrounding comments do not warn about that data handling.

Missing User Warnings

Medium

• Confidence: 88% confidence
• Finding: The function stores generated replies keyed by user_message via cache_reply, which is a file/data write involving user-provided content. While comments mention caching for implementation purposes, there is no user-facing warning that chat content may be retained locally, which is relevant to privacy and data handling.

Natural-Language Policy Violations

Medium

• Confidence: 86% confidence
• Finding: The prompt instructions are written to produce replies in Chinese and use Chinese-specific audience forms, but there is no indication that this locale restriction is optional or user-selected. Under the policy, forcing a specific language without opt-in is a natural-language policy concern unless the locale constraint is clearly documented and justified.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The code forwards `user_name` and `content` from live chat to `generate_reply`, which implies transmission of user-generated data to an external AI service. Although the action is commented in code, there is no confirmation prompt or user-facing disclosure to affected users before their chat content is processed externally.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: This block persists timestamps, usernames, user messages, and AI replies to `REPLY_FILE`. While the code performs the write directly, it does not provide any disclosure to users that their messages are being stored locally, creating an undisclosed data-retention behavior.

Natural-Language Policy Violations

Low

• Confidence: 88% confidence
• Finding: The parameters force `browser_language` to `zh-CN` and `tz_name` to `Asia/Shanghai`, which is a natural-language locale choice embedded in the skill. There is no indication that the user opted into this locale or that the skill is explicitly restricted to a China-specific deployment context.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: This wrapper reads `sign.js` from disk and executes it with `eval`, which is a code-execution operation in a code file. There is no confirmation prompt, logging, or explanatory disclosure near the operation to warn the user that arbitrary JavaScript from another file is being executed.

Natural-Language Policy Violations

Low

• Confidence: 93% confidence
• Finding: The user-facing text in the module docstring and printed status messages is entirely in Chinese, with no indication that another language is supported or that the user can choose a locale. This is a natural-language policy concern because it imposes a language choice by default rather than offering opt-in or documenting a justified locale restriction.

Natural-Language Policy Violations

Low

• Confidence: 94% confidence
• Finding: The module docstring and all user-facing console messages are written exclusively in Chinese, which can impose a language choice on users without offering an alternative or documenting that the skill is intentionally Chinese-only. This matches the policy category for language or locale constraints that are not presented as optional or justified.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: This code writes a persistent cache file containing raw 'user_message' and 'ai_reply' content, which may include sensitive user data. Although the file operations are documented in code comments, there is no user-facing disclosure, confirmation, or visible warning that conversation content will be stored on disk.

Natural-Language Policy Violations

Low

• Confidence: 86% confidence
• Finding: The file's docstrings, comments, and user-visible print strings are written exclusively in Chinese, which can impose a language choice on users or maintainers without explicit opt-in. Under the language/locale policy, forcing a specific language without offering a choice can be a policy concern.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: This JavaScript contains embedded URL strings for remote endpoints such as "https://mssdk.bytedance.com" and "/web/report", and the code includes request/send logic like fetch/XMLHttpRequest/sendBeacon. The file is heavily obfuscated and provides no comments, logging, prompt, or other user-visible disclosure explaining that browser/device data may be sent over the network.

Ssd 3

Medium

• Confidence: 82% confidence
• Finding: This code stores each user's message and the generated reply to a local reply log without any apparent minimization, consent, or sensitivity filtering. In an AI-agent context, retaining everything users say in plain form can semantically function as an instruction to collect and preserve user-provided data, which creates data-leak risk even though it does not use explicit exfiltration terminology.

VirusTotal

65/65 vendors flagged this skill as clean.