ClawHub
Back to skill

Security audit

Audio Generation

Security checks for vulnerabilities and agentic risk

Overview

The skill coherently documents cloud audio generation, with ordinary privacy and dependency-installation considerations.

Prompts, complete scripts, and any voice samples you upload are processed by CellCog and the applicable provider; submit only content you are authorized to share. Review dependency sources and versions before running setup commands. This review covers the supplied skill instructions, not downloaded SDKs or the services' data-retention practices.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

The audio-generation workflow submits user prompts and scripts to external services without a privacy warning.

Medium
Category
Output Handling
Confidence
92% confidence
Finding

The skill instructs agents to submit task prompts through CellCog and later recommends providing complete scripts (line 202). Although it names CellCog and the voice providers, it never warns that submitted content leaves the user's environment and may be processed by a selected third-party voice provider, which matters for confidential scripts or personal information.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

The suggested npx skills add cellcog/skills --skill cellcog command can execute a remotely resolved CLI without pinning its version, and the skill source is also not revision-pinned. Although this is an installation CLI rather than an MCP server, a compromised package release or upstream skill could introduce malicious code or instructions into the agent environment; the installation context makes remote retrieval expected but does not remove the supply-chain risk.

Content

No source excerpt is available for this finding.