ClawHub
Back to skill

Security audit

Funasr Asr

Security checks for vulnerabilities and agentic risk

Overview

This appears to be a media transcription skill whose web-download and FFmpeg behavior matches its purpose, but users should only use trusted media URLs.

Install only if you want a transcription helper that may download remote media. Use trusted http/https URLs or local files, avoid private/internal URLs, and review the helper script’s URL handling if running it in a shared or sensitive network.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

subprocess module call

Medium
Category
Dangerous Code Execution
Content
# 回退到 ffmpeg
            try:
                cmd = ["ffmpeg", "-i", url, "-c", "copy", output_path, "-y"]
                subprocess.run(cmd, check=True, capture_output=True)
                print(f"✅ 视频已保存: {output_path}")
                return output_path
            except subprocess.CalledProcessError as e:
Confidence
82% confidence
Finding
Although shell injection is avoided, ffmpeg is invoked on a user-controlled URL, and ffmpeg supports multiple protocols and remote inputs. In a skill context where arbitrary users may supply URLs, this can enable SSRF-style network access to internal services or retrieval of local/unsupported resources depending on ffmpeg build configuration and runtime environment.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The trigger description is broad enough to activate on common speech-to-text and media transcription requests, including "download video from web and transcribe," which can cause the skill to run in situations the user did not explicitly intend. Over-broad activation increases the chance of unnecessary local processing, unintended network access, and privacy-sensitive transcription of user media.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill advertises downloading web video for transcription but does not warn that this may contact external services, transfer user-supplied URLs, consume bandwidth, or process potentially sensitive content. In an agent environment, omitting this disclosure can lead to unexpected network activity and privacy exposure, especially if users assume the skill is fully local.

Static analysis

Detected: suspicious.dangerous_exec

Shell command execution detected (child_process).

Critical
Code
suspicious.dangerous_exec
Location
index.js:36