ClawHub
Back to skill

Security audit

podcast-transcribe

Security checks across malware telemetry and agentic risk

Overview

The skill mostly does what it says, but it under-discloses a referral signup link, automatic episode filtering, and paid third-party transcription data flows.

Review this before installing if you handle private feeds, licensed audio, or sensitive speech. Use a neutral AuralWise signup URL if referral tracking is unwanted, keep the API key out of shared repositories, start with test or download-only mode, and verify the script's episode filtering will not skip content you expect to process.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

Lp3

Medium
Category
MCP Least Privilege
Confidence
96% confidence
Finding
The skill instructs the agent to read environment-related configuration, read and write local files, and make network requests, but it does not declare corresponding permissions. Undeclared capabilities reduce transparency and can cause the agent to access local secrets, modify files, or contact external services without clear user or platform-level consent.

Tp4

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding
The skill behavior diverges from its stated purpose by embedding a referral-style signup link and, per the finding, applying episode-title filtering not disclosed in the description. Hidden monetization or undisclosed processing logic undermines user trust and can manipulate user actions or silently exclude data, which is especially risky in an automation skill handling external downloads and paid API usage.

Missing User Warnings

Medium
Confidence
92% confidence
Finding
The documentation explicitly instructs users to provide a publicly accessible audio URL to a third-party transcription service, but it does not warn that podcast audio, feed-derived URLs, and related metadata will be disclosed to that external provider. In this skill context, that omission matters because the workflow is centered on downloading and transcribing media at scale, which can expose private or licensed audio sources, access patterns, and content to AuralWise without informed consent or review.

External Transmission

Medium
Category
Data Exfiltration
Content
## API 端点

Base URL: `https://api.auralwise.cn/v1`

### 1. 查询账户信息
Confidence
81% confidence
Finding
This file documents use of an external API endpoint, which means account data, task metadata, audio references, and transcription results are transmitted outside the local environment. In a transcription skill, external transmission is expected, but it is still security-relevant because the service processes user-supplied media and billing/account information, increasing exposure to third-party data handling risks.

External Transmission

Medium
Category
Data Exfiltration
Content
for attempt in range(1, max_retries + 1):
            try:
                resp = requests.post(
                    f"{self.BASE_URL}/tasks",
                    headers=self.headers,
                    json=payload,
Confidence
97% confidence
Finding
The script transmits podcast audio URLs and associated metadata to a third-party service (AuralWise) for transcription. This is security-relevant because episode content, potentially sensitive speech, and listening/transcription activity leave the local environment and are processed by an external provider, which may violate user expectations or organizational data-handling rules if not clearly consented to.

External Transmission

Medium
Category
Data Exfiltration
Content
class AuralWiseTranscriber:
    """AuralWise API 客户端"""

    BASE_URL = "https://api.auralwise.cn/v1"

    # 默认转写选项（播客单人主讲场景）
    DEFAULT_OPTIONS = {
Confidence
96% confidence
Finding
The hard-coded external API endpoint establishes that the skill is designed to send data off-host to AuralWise. In the context of podcast transcription this is expected, but it still creates a real data exfiltration surface for audio content and metadata, especially if users assume processing is local or if feeds include non-public content.

VirusTotal

61/61 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

Detected: suspicious.exposed_secret_literal

File appears to expose a hardcoded API secret or token.

Critical
Code
suspicious.exposed_secret_literal
Location
scripts/podcast_pipeline.py:825

File appears to expose a hardcoded API secret or token.

Critical
Code
suspicious.exposed_secret_literal
Location
SKILL.md:38