ClawHub
Back to skill

Security audit

Audio Stream Upload

Security checks for vulnerabilities and agentic risk

Overview

This appears to be an audio upload skill, but it needs review because it can batch-send local audio files, metadata, and API keys to a third-party service while declaring broader triggers and tool authority than necessary.

Install only if you trust the third-party upload service and intend to send the selected audio files and metadata there. Use dedicated API keys with limited permissions, avoid sensitive or copyrighted media unless you have authorization, and require the agent to confirm exact file paths and destination before any upload or shell command runs.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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)

Intent-Code Divergence

High
Confidence
95% confidence
Finding
The documented inputs and outputs describe a generic content-processing skill rather than a narrowly scoped audio upload tool, which creates capability confusion. In an agent setting, misleading I/O contracts can cause the skill to be invoked on unintended data and may lead users or orchestrators to pass sensitive content that is then handled or transmitted in ways they did not expect.

Intent-Code Divergence

Medium
Confidence
91% confidence
Finding
The documentation advertises command-execution capability even though the examples primarily implement file and network upload behavior. This mismatch expands the perceived authority of the skill and may cause agents or users to authorize shell execution unnecessarily, increasing the risk of unsafe local actions or misuse of the exec tool.

Vague Triggers

Medium
Confidence
84% confidence
Finding
The activation description is broad and ambiguous, including unrelated media-processing scenarios beyond simple audio upload. Overbroad triggering conditions can cause an agent to select this skill in contexts where file transmission, metadata handling, or local tool use was not intended, increasing the chance of inappropriate data exposure or unsafe execution paths.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The skill includes external upload behavior and mentions command execution without a clear up-front warning that audio files, metadata, and credentials are transmitted to a third-party service. In agent workflows, lack of explicit disclosure undermines informed consent and can result in sensitive or proprietary media being exfiltrated outside the local environment.

External Transmission

Medium
Category
Data Exfiltration
Content
create_data = {'title': title, 'type': 'audio'}
            if config:
                create_data.update(config)
            resp = requests.post(
                f'{self.base_url}/videos/create',
                headers={**self.headers, 'Content-Type': 'application/json'},
                json=create_data
Confidence
98% confidence
Finding
The code performs external HTTP requests to a non-standard third-party domain and sends audio metadata along with authentication headers, followed by file upload. In this skill context, external transmission is expected functionality, but it is still dangerous because users may not realize that local files, metadata, and API secrets are being sent off-platform to an unvetted endpoint.

External Transmission

Medium
Category
Data Exfiltration
Content
def upload_large_file(self, file_path, title, chunk_size=10*1024*1024):
    """分片上传大文件，支持断点续传"""
    file_size = os.path.getsize(file_path)
    resp = requests.post(f'{self.base_url}/videos/create',
headers, 'Content-Type': 'application/json'},
        json={'title': title, 'type': 'audio'})
    audio_id = resp.json()['data']['id']
Confidence
97% confidence
Finding
The large-file upload flow also transmits data to the same external service, including metadata and file chunks, which can expose sensitive media contents at scale. Because this path supports chunked transfer and resume logic, it may facilitate bulk exfiltration of large local files if invoked inappropriately or without strong user awareness.

Static analysis

No suspicious patterns detected.