Security audit

Career

Security checks across malware telemetry and agentic risk

Overview

This skill provides career advice and keeps a clearly scoped local career profile, which is privacy-sensitive but disclosed and aligned with its purpose.

Install only if you are comfortable with the agent keeping a local career profile under ~/Clawic/data/career/. That profile may include compensation, constraints, values, visa-related status, targets, and coaching preferences, so review or delete those files if you do not want that history retained.

SkillSpector

By NVIDIA

Vulnerability Patterns

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
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (5)

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: The skill instructs the agent to create and maintain persistent user profile files containing compensation history, constraints, values, and targets. That expands behavior from one-time career advice into longitudinal storage of sensitive personal data, which increases privacy and misuse risk if users are not clearly informed and have not consented.

Description-Behavior Mismatch

Medium

Confidence: 97% confidence
Finding: Recording agent-observed behavioral patterns in a persistent memory file goes beyond explicit user-provided career facts and creates a profile based on inference. This is riskier because inferred preferences and decision traits may be inaccurate, sensitive, and stored without the user’s awareness, creating privacy and trust issues.

Natural-Language Policy Violations

Medium

Confidence: 89% confidence
Finding: The file explicitly states that legal guidance defaults to US assumptions unless a `region` config is set, which can cause the agent to give jurisdiction-specific advice without first confirming the user's location. In a career/legal-adjacent context, this is risky because severance, notice, unemployment, and visa rules vary materially by country, and wrong defaults could mislead a user into missing deadlines or rights.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The setup directs the agent to read and create persistent files for sensitive career information without any user-facing notice about storage, retention, or reuse. Because career data can include compensation, constraints, and potentially immigration-related status, silent persistence materially increases privacy risk even if the overall purpose is legitimate.

Missing User Warnings

Medium

Confidence: 98% confidence
Finding: The recording rules instruct ongoing updates to persistent memory based on user disclosures and agent observations, but provide no warning or consent mechanism for continued retention. In the context of a career advisory skill, this can accumulate a detailed and sensitive history over time, making the absence of transparency more dangerous rather than less.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.