Back to skill

Security audit

Speech Recognition

Security checks for vulnerabilities and agentic risk

Overview

This skill is a straightforward speech-to-text setup guide, with ordinary dependency and model-download risks that users should review before running.

Before installing, use a dedicated virtual environment, pin reviewed package versions, and confirm that downloading Systran/faster-whisper-base from HuggingFace is allowed by your network and data policies. Non-Chinese-speaking users should translate the instructions before running commands.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (1)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:35
Finding

Unpinned Third-Party Dependency Installation

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 35–38
Vulnerability Type: Unpinned third-party dependencies
Risk Level: Medium

bash
source /opt/conda/bin/activate py314
pip install pysilk faster-whisper

Technical Analysis

The installation instructions request pysilk and faster-whisper without specifying reviewed versions or package hashes. Consequently, dependency resolution can change over time and select future, compromised, or otherwise unreviewed releases. Python package installation may execute package-controlled build or installation logic, and installed packages execute code when imported.

The use of a shared, fixed environment at /opt/conda/bin/activate py314 may also increase the affected scope compared with an isolated environment dedicated to this skill. The audit found no evidence that either named package is currently malicious; the finding concerns unsafe dependency-management practices.

Attack Path

  1. An attacker compromises a package publisher account, package release, or relevant supply-chain component.
  2. The attacker publishes a malicious version that still satisfies the unconstrained package request.
  3. A user follows the documented pip install pysilk faster-whisper command.
  4. pip resolves and retrieves the attacker-controlled distribution.
  5. Malicious code executes during package installation, build processing, or later package import.
  6. The code operates with the permissions of the user running pip or the transcription workflow.

Impact Assessment

Successful exploitation could permit arbitrary code execution with the invoking user's privileges. Potential effects include access to files and credentials available to that user, modification of the shared Python environment, tampering with speech-recognition output, and execution during later uses of the compromised dependency. No privilege escalation beyond the invoking user's existing permissi ...[truncated 40 chars]

Remediation
View remediation

Remediation Suggestions

  • Pin every direct and transitive dependency to an exact, reviewed version.
  • Maintain a lock file and require package hashes, such as with pip install --require-hashes.
  • Install dependencies into a dedicated, least-privilege virtual environment rather than a shared environment.
  • Retrieve packages only from an explicitly configured, trusted package index.
  • Scan and review resolved distributions before deployment.
  • Establish a controlled update process that tests and approves dependency changes before updating pins and hashes.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (2)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The natural-language content of the skill is presented only in Chinese, which can amount to a language policy violation when no user opt-in or alternative language is provided. The file does not indicate that the skill is intentionally region-specific or that users may choose another language.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
81% confidence
Finding

This markdown file instructs the user to download a model from HuggingFace, which involves a network operation, but it does not explicitly warn that the command will contact an external service. Under the markdown-specific warning criteria, network-related behavior that may affect privacy or system integrity should be disclosed.

Content

No source excerpt is available for this finding.