Meta Knowledge Base

• Outcome: Review
• Latest audit: May 29, 2026 at 9:27 PM UTC
• Version:
  1.0.0

Overview

This is a coherent local knowledge-base skill, but it promotes broad automatic capture and persistent storage of private content without clear controls.

Install only if you intentionally want a persistent personal knowledge base. Use explicit source selection, avoid secrets and private folders, and verify how to inspect and remove data under ~/.meta_knowledge before enabling conversation hooks, watched folders, email ingestion, or automatic learning.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (5)

Missing User Warnings

High

• Confidence: 97% confidence
• Finding: The skill advertises automatic capture of chats, documents, web pages, and monitored folders without any visible consent flow, scope limits, or privacy warning. Because these sources commonly contain personal, confidential, or credential-bearing data, automatic ingestion can lead to overcollection and persistent storage of sensitive information.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: Email ingestion is especially sensitive because emails often contain personal data, internal business information, tokens, attachments, and password reset links. Documenting email parsing with no warning or safeguards encourages unsafe handling of highly sensitive content.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: The OpenClaw hook example automatically stores derived content from user messages after every interaction, without notice or consent. In practice, conversation content can include secrets, personal information, proprietary prompts, or regulated data, so silent persistence materially raises privacy and leakage risk.

Ssd 3

Medium

• Confidence: 94% confidence
• Finding: Unbounded storage of user interaction content creates a natural-language data leakage path: later search, RAG retrieval, summaries, or exports may surface prior sensitive conversations to unintended users or contexts. The skill's knowledge-base purpose makes this risk more acute because retained content is designed to be retrieved and reused.

Ssd 3

Medium

• Confidence: 91% confidence
• Finding: The phrase 'Remember everything' promotes indiscriminate retention behavior, which is unsafe for a system intended to ingest conversations, documents, web content, and emails. In this context, broad retention increases the chance that sensitive or regulated information is stored longer than necessary and later exposed through retrieval features.

VirusTotal

65/65 vendors flagged this skill as clean.