Back to skill

Security audit

whisperx

Security checks for vulnerabilities and agentic risk

Overview

This skill is a straightforward guide for generating word-level audio timestamps with Replicate WhisperX, with a visible supply-chain caution around unpinned install commands.

Before installing, verify that you trust the PrunaAI skill source and prefer a pinned or otherwise vetted installer when possible. Use this skill only for audio you are comfortable sending to Replicate, and provide the Replicate token through normal environment handling rather than pasting secrets into shared logs.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (9)

Rp1

Medium
Category
MCP Rug Pull
Confidence
93% confidence
Finding

The skill instructs users to execute npx skills add ... without pinning the skills package/tool version. Because npx resolves and runs the latest available package by default, a compromised or breaking upstream release could execute unintended code in the agent environment during installation.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
93% confidence
Finding

This line again relies on npx skills without a pinned version, creating a supply-chain execution risk. If the package or one of its resolution paths is hijacked or updated maliciously, the agent may run arbitrary code simply by following the skill instructions.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
93% confidence
Finding

The unpinned npx skills invocation can fetch and execute mutable remote code at runtime. In an agent skill context, that is especially risky because the instructions are likely to be followed automatically or semi-automatically during workflow setup.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding

This recommendation to install another skill via unpinned npx skills exposes users to the same mutable dependency execution risk. An attacker controlling or poisoning the package resolution path could achieve code execution in the local environment.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding

The skill points users to an unpinned npx skills command, which means the executed installer/runtime can change over time without review. That creates a classic supply-chain weakness with potential arbitrary code execution if the upstream package is compromised.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding

This line continues the pattern of invoking a mutable npx target without version pinning. Because installation commands are high-trust actions, any compromise of the package ecosystem or namespace could lead to immediate code execution on the host.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

The skill's follow-on recommendation uses npx skills without locking the executed version, preserving the same supply-chain risk later in the workflow. In chained skill ecosystems, repeated unpinned installs broaden the attack surface and make compromise more likely over time.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

This unpinned npx skills command can execute whatever version is current at resolution time, which is unsafe for reproducible and trusted agent workflows. If exploited, it could result in arbitrary code execution, credential theft, or tampering with subsequent workflow steps.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

The final referenced npx skills command is also unpinned, making it another true supply-chain vulnerability. Because the skill also references API tokens and external service usage, compromise of the installer path could expose sensitive credentials or alter downstream behavior.

Content

No source excerpt is available for this finding.