Back to skill

Security audit

Engineering-Build

Security checks for vulnerabilities and agentic risk

Overview

The skill’s engineering, deployment, and automation capabilities are disclosed and supported by approval, credential protection, and rollback safeguards.

Before cloud or scheduled automation work, specify the target project, permitted changes, spending limits, storage retention, and schedule. Use narrowly scoped credentials and review deployment plans and rollback steps before approving a release.

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)

Executable deployment pipeline lacks warnings about changes to cloud resources and running services.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

L138 instructs the agent to execute the scenario task chains in sequence, making these workflows more than a passive task catalog. The deployment chain at L144 includes cloud provisioning, storage permissions and lifecycle configuration, and deployment (defined at L41–L45), but provides no warning about potential charges, data-access changes, or service disruption.

Content

No source excerpt is available for this finding.