Back to skill

Security audit

Preserve Human Code Edits - load human changes into context before editing

Security checks for vulnerabilities and agentic risk

Overview

This skill provides a simple Git workflow for preserving human edits, with a disclosed but caution-worthy instruction to always commit changes after editing.

Install only if you want the agent to treat commits as part of its normal editing workflow. Before using it on sensitive or shared repositories, tell the agent whether commits require your confirmation and ask it to show the diff and proposed commit message first.

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
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill instructs the agent to always commit changes after edits without requiring explicit user confirmation or presenting a warning. This can cause unintended repository state changes, persist sensitive or partial modifications, and violate user expectations in environments where the agent should not perform irreversible source-control actions autonomously.

Content

No source excerpt is available for this finding.