Back to skill

Security audit

Tutoring

Security checks for vulnerabilities and agentic risk

Overview

The skill coherently uses CellCog for tutoring, with ordinary privacy and dependency-installation considerations.

Prompts and included essays, code, or other materials are sent to CellCog. Share only material you intend the service to receive, protect your API key, and review the separate CellCog dependency before installation. This review covers the supplied skill instructions, not downloaded dependencies or the service's data handling.

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

Medium
Category
Output Handling
Confidence
96% confidence
Finding

The usage instructions direct the agent to submit the user's task prompt to the third-party CellCog service through client.create_chat. Although CellCog is named, the skill does not warn that submitted prompts—including any personal information, essays, or proprietary code they contain—leave the user's environment.

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 setup command uses npx skills without a version, allowing installation and execution of a changing npm package with the invoking user's permissions. A compromised future release could execute malicious code during setup; however, the finding's MCP-server characterization is inaccurate, and the supplied file shows no evidence of an actual compromise.

Content

No source excerpt is available for this finding.