Security audit

ecommerce-golive-check

Security checks for vulnerabilities and agentic risk

Overview

This store-launch checklist needs review because it can approve incomplete checks and lacks clear limits for tests that may affect real orders.

Use an isolated test store, payment-provider test mode, and a controlled inbox; explicitly approve any production testing. Independently verify every required check and waiver before accepting launch approval. The static scan reported no findings, but direct testing confirmed the gate weaknesses described above.

A.I.G

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

SkillSpector

Vulnerability Patterns

Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
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

Findings (3)

The blocking launch gate can certify readiness without any checklist evidence.

High

Category: Excessive Agency
Confidence: 99% confidence
Finding: The gate treats the absence of recognized blocking failures as proof of readiness, without validating required checklist coverage or allowed status and severity values. An empty object therefore produces LAUNCH-READY and exit code 0, while omitted checks or unrecognized statuses can also bypass the gate, potentially approving a release whose payment, authorization, or compliance controls were never verified.
Content: No source excerpt is available for this finding.

Real-order and signed-webhook tests lack a warning about live-store side effects.

Medium

Category: Output Handling
Confidence: 94% confidence
Finding: The workflow requires checkout, signed-webhook replay, and real-order email testing without establishing a sandbox boundary or requiring authorization for production mutations. Because this skill also applies to existing live stores, following it could cause real charges, duplicate order processing, inventory changes, fulfillment, or unintended customer notifications; delegated skills might provide safeguards, but none are established here.
Content: No source excerpt is available for this finding.

Real-order and inbox testing is instructed without a warning about live side effects.

Medium

Category: Output Handling
Confidence: 94% confidence
Finding: The catalog directly instructs a real signed webhook replay and real test-order delivery without specifying an isolated environment, controlled recipients, or authorization for live side effects. Against a production store, these operations can trigger payment-state changes, duplicate downstream processing, fulfillment, or customer emails; testing actual delivery is legitimate, but does not require unrestricted production activity.
Content: No source excerpt is available for this finding.