ClawHub
Back to skill

Security audit

Groq Voice Transcribe

Security checks across malware telemetry and agentic risk

Overview

This is a straightforward Groq cloud transcription skill, with privacy-sensitive behavior that is disclosed and aligned with its purpose.

Install only if you are comfortable sending selected audio files to Groq for cloud transcription. Use a dedicated, revocable Groq API key, check Groq's privacy and retention terms for your use case, and avoid transcribing recordings that contain secrets or sensitive conversations unless that processing is acceptable.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Rogue AgentSelf-Modification, Session Persistence
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
94% confidence
Finding
The skill documents execution of a local shell script (`{baseDir}/scripts/transcribe.sh`) but does not declare corresponding permissions. This creates a trust and review gap: an agent or user may authorize the skill based on incomplete metadata, while the skill still drives shell-capable behavior that can access local files and transmit audio to a remote API.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The script uploads the provided audio file to Groq's external speech-to-text API, but it gives no user-facing disclosure, consent prompt, or privacy warning before transmitting potentially sensitive voice data off-device. In the context of an agent skill that may automatically process user voice messages, this creates a real privacy and data-handling risk because users may reasonably expect local processing or may not realize third-party cloud transmission is occurring.

External Transmission

Medium
Category
Data Exfiltration
Content
mkdir -p "$(dirname "$out")"

curl -sS https://api.groq.com/openai/v1/audio/transcriptions \
  -H "Authorization: Bearer $GROQ_API_KEY" \
  -H "Accept: application/json" \
  -F "file=@${in}" \
Confidence
92% confidence
Finding
This code makes an outbound request to https://api.groq.com/openai/v1/audio/transcriptions and includes the full audio file in the multipart form body, which is an intentional external data transfer. While the transmission is expected for the skill's purpose, it is still a genuine security/privacy concern because voice recordings can contain sensitive personal, financial, or authentication information, and the skill context makes the absence of disclosure and consent more dangerous.

Session Persistence

Medium
Category
Rogue Agent
Content
Get one from:

- https://console.groq.com/
- Sign in, open **API Keys**, then create a key

## Easiest setup in OpenClaw
Confidence
83% confidence
Finding
The skill instructs users to give the assistant their Groq API key and have it write the secret into persistent local configuration (`~/.openclaw/openclaw.json`). Persisting credentials via conversational flow increases the chance of secret exposure through logs, prompt history, agent memory, or overbroad file access by other tools and skills.

VirusTotal

56/56 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.