Back to skill

Security audit

PDF Document Generation

Security checks for vulnerabilities and agentic risk

Overview

The skill coherently uses CellCog to generate requested documents, with ordinary data-sharing and dependency-installation considerations.

Prompts and supplied document content go to CellCog, so share only information you are authorized to disclose. Review the separate CellCog SDK and base skill before enabling them, and prefer reviewed, pinned dependency versions. The scanner reported complete coverage, and independent inspection covered the entire supplied artifact; external dependency behavior and service data retention remain outside this review.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

The workflow sends document content to CellCog without an explicit privacy warning.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

The workflow submits the user's task prompt to CellCog through client.create_chat, while the examples and tips encourage including personal work histories, business metrics, and contract details. Although CellCog is named as the provider, the skill does not explicitly warn that supplied content is transmitted to a third-party service.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
84% confidence
Finding

The installation command invokes an unpinned npm package through npx and selects skill content without an immutable revision, allowing the code and instructions retrieved to change over time. A compromised upstream release could execute code with the installing user's permissions or introduce malicious agent instructions. This is a conditional supply-chain exposure, not evidence of an existing compromise; the static finding incorrectly describes the command as an MCP server reference.

Content

No source excerpt is available for this finding.