学习日志系统

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:35 PM UTC
• Version:
  1.0.0

Overview

This skill is a disclosed local learning journal that writes notes to workspace files, with no evidence of network transfer, credential access, or hidden behavior.

Install only if you want the agent to keep local plaintext learning notes. Avoid saving secrets, credentials, customer data, private incident details, or sensitive paths, and set the intended workspace when using the helper script so entries are written where you expect.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The trigger conditions are extremely broad—task completion, failures, discoveries, decisions, and workflow optimizations—which can cause the skill to activate during many ordinary interactions. In a journaling skill that writes persistent files, overbroad activation materially raises the risk of capturing unnecessary or sensitive context without a deliberate user request.

Ssd 3

Medium

• Confidence: 96% confidence
• Finding: The skill instructs the agent to automatically persist learned information from tasks and conversations into long-lived workspace memory files. Without sensitivity filtering, consent, or data-minimization rules, this can store secrets, proprietary prompts, internal decisions, user data, or other sensitive context in plaintext where later tools, users, or processes may access it.

Ssd 3

Medium

• Confidence: 95% confidence
• Finding: The examples and write-timing guidance encourage broad logging of task context, errors, discoveries, decisions, and workflow changes, but provide no safeguards for sensitive material. That combination strongly increases the chance that the agent will save confidential conversational content or operational details into durable notes, creating persistent data leakage and secondary exposure risk.

VirusTotal

67/67 vendors flagged this skill as clean.