Security audit

B2A

Security checks for vulnerabilities and agentic risk

Overview

This is a documentation-only skill about making products usable by AI agents; it is coherent and non-executable, though users should add stronger consent and privacy safeguards before implementing the commerce patterns.

Safe to install as guidance material. Before implementing its autonomous purchasing or tracking examples, require explicit user authorization, spending and quantity caps, clear notifications, easy cancellation, scoped credentials, and privacy controls for user and agent identifiers.

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

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The skill explicitly recommends tracking per-query analytics tied to an `agent_id`, selection status, and conversion outcomes, but provides no privacy boundaries, retention limits, minimization guidance, or disclosure expectations. This creates a realistic risk of unnecessary profiling, cross-session correlation, and collection of identifiable or pseudonymous agent/user behavior data beyond what is needed for service operation.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill promotes autonomous reordering ('No human decision point') without any guardrails around user consent, approval thresholds, cancellation, or clear disclosure that purchases may occur automatically. In a retail/ecommerce context this increases the risk of unauthorized or unexpected purchases, especially if downstream agents implement the pattern literally.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The subscription API example includes persistent identifiers like user_id and agent_id and creates recurring purchasing behavior, but it omits privacy, consent, and authorization safeguards. Without guidance on minimization, authentication, scoped tokens, and user approval for subscription creation, implementers could expose personal data and enable unauthorized recurring orders.

Static analysis

No suspicious patterns detected.