Back to skill

Security audit

Finetune

Security checks for vulnerabilities and agentic risk

Overview

This skill is a simple Finetune reference helper that prints local documentation and does not access credentials, network services, or persistent system settings.

This appears safe to install as a local reference skill. Expect static documentation output only; be aware that some command text is slightly inconsistent and should not be treated as authoritative operational guidance without checking real Finetune documentation.

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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/script.sh (reported line 64)May include surrounding context.

sh
## Prerequisites
- Basic understanding of devtools concepts
- Required tools and access credentials
- System meeting minimum requirements

## Installation

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
97% confidence
Finding

The help output claims to show a versioned banner, but the here-document is single-quoted, so $VERSION is not expanded and users see the literal string rather than the real version. This is a documentation-to-code contradiction in the displayed interface text, even though the underlying script version variable is set correctly.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
99% confidence
Finding

The cheatsheet says troubleshooting is an available command, but the command dispatcher only implements debugging and will reject troubleshooting as unknown. This actively misstates the skill's interface and can mislead users about what the tool actually supports.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.