Back to skill

Security audit

object_counter

Security checks for vulnerabilities and agentic risk

Overview

This skill is a small local image object-counting helper with no evidence of hidden network access, persistence, credential use, or destructive behavior.

Installers should understand that this skill runs local image processing code on files they provide. The main issue is polish: the script advertises two unsupported tool names, so users should use only --tool count unless the publisher updates the interface.

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
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Description-Behavior Mismatch

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest describes an object-counting skill, and the module docstring also presents it as a counting library, but the argument parser exposes tool choices for 'denoise' and 'super_resolution' in addition to 'count'. In practice only 'count' is implemented, so the exposed interface claims broader behavior than the code actually supports.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
97% confidence
Finding

The parser is described as 'a set of image processing tools' and offers 'denoise' and 'super_resolution', but the dispatch logic only handles 'count' and raises NotImplementedError for anything else. This is an active contradiction between the code's inline interface/documentation and its actual behavior.

Content

No source excerpt is available for this finding.