Security audit

Ecommerce

Security checks across malware telemetry and agentic risk

Overview

This ecommerce skill stores local operational notes and has privacy guardrails; the scanner concerns are real cautions but are mostly disclosed and purpose-aligned.

Before installing, be comfortable with the skill keeping local ecommerce memory under ~/Clawic/data, including business contacts, supplier/wholesale terms, metrics, deadlines, and incident summaries. Do not use its legal, tax, email, SMS, or consumer-law guidance without checking the current rules for your jurisdiction and ensuring marketing consent, opt-out handling, and retention practices are correct.

SkillSpector

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 (5)

Natural-Language Policy Violations

Medium

Confidence: 88% confidence
Finding: Defaulting legal and fiscal behavior from `profile.yaml` or assumptions creates a hidden trust boundary: shared profile data or unstated assumptions can drive compliance recommendations without the user's active approval in the current session. In a skill that covers taxes and consumer law, this increases the chance of incorrect registration, disclosure, or filing advice.

Natural-Language Policy Violations

Medium

Confidence: 88% confidence
Finding: Defaulting legal and fiscal behavior from `profile.yaml` or assumptions creates a hidden trust boundary: shared profile data or unstated assumptions can drive compliance recommendations without the user's active approval in the current session. In a skill that covers taxes and consumer law, this increases the chance of incorrect registration, disclosure, or filing advice.

Missing User Warnings

Low

Confidence: 88% confidence
Finding: The skill explicitly instructs the agent to write account terms, named contacts, tax numbers, credit limits, and related business data into shared memory files without any user-facing consent prompt, minimization guidance, or data-handling boundary. While this appears operationally motivated rather than malicious, it can cause unnecessary propagation of sensitive commercial and personal data into shared stores, increasing privacy, confidentiality, and retention risk.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The skill gives operational guidance for abandoned-cart email and SMS sequences but omits any warning that these messages must be conditioned on lawful consent, jurisdictional marketing rules, and platform policies. In an ecommerce skill, this omission is more dangerous because users are likely to implement the workflow directly, creating risk of unlawful marketing outreach, complaints, spam filtering, or regulatory exposure.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The document tells operators to write support outcomes into metrics, pain points, and incident files, which creates a real risk that customer-specific details are copied into persistent markdown records outside controlled support systems. Although it earlier says transcripts and identities may not be stored, the later instruction is broad enough that staff or downstream agents could still record incident details containing personal data, increasing privacy, retention, and unauthorized disclosure risk.

VirusTotal

62/62 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

Detected: suspicious.exposed_secret_literal

File appears to expose a hardcoded API secret or token.

Critical

Code: suspicious.exposed_secret_literal
Location: memory-template.md:73