Back to skill

Security audit

Wireframe

Security checks for vulnerabilities and agentic risk

Overview

The skill’s CellCog integration fits its prototype-building purpose, with privacy and installation considerations.

Use content you are authorized to send to CellCog, and avoid confidential material in shareable prototypes. Check hosting access and retention terms, and review or pin installation dependencies. The bundled instructions were fully inspected; the external SDK and prerequisite skill were not included.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • 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
Findings (2)

The workflow sends user task content to CellCog without a privacy warning.

Medium
Category
Output Handling
Confidence
93% confidence
Finding

The usage workflow submits the user's task prompt to the third-party CellCog service, but the skill does not explicitly warn that the supplied content leaves the user's environment. This omission is particularly relevant because line 262 encourages using real content, which could include confidential product details or user data.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
86% confidence
Finding

The recommended npx skills add command can download and execute an unpinned npm package, while the skill source also lacks an immutable revision. A compromised future release could execute code with the installing user's permissions or introduce malicious agent instructions. This is a supply-chain reproducibility weakness, not an MCP server configuration issue as the finding's label suggests; exploitation requires following the optional installation instructions and a compromised upstream dependency.

Content

No source excerpt is available for this finding.