Back to skill

Security audit

Taste

Security checks for vulnerabilities and agentic risk

Overview

This testing skill directs your agent to discard code without requiring a backup, checking ownership, or asking first.

Before using this skill, preserve existing work and require deletion to be limited to explicitly approved changes. Its mandatory discard rules should be replaced with a recoverable workflow that respects your instructions.

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 (1)

The skill directs deletion of implementation code without warning the user about potential loss of existing work.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

The skill unconditionally instructs the agent to delete implementation written before tests, without checking ownership, preserving uncommitted changes, or obtaining user approval. Although the purpose is legitimate test-driven development, repeated demands to discard work—including hours of implementation—could cause loss of user-authored or otherwise unrecoverable changes when applied to an existing workspace.

Content

No source excerpt is available for this finding.