Learnings Skill

Security checks across malware telemetry and agentic risk

Overview

This skill does what it claims: it stores user-directed learning notes in a MeiliSearch index, but users should treat that index as potentially containing sensitive workflow history.

Install only if you are comfortable keeping a persistent searchable history of mistakes, fixes, preferences, and selected memory-note content. Keep MEILI_HOST pointed at a trusted local or controlled MeiliSearch instance, protect the MEILI_KEY, review dry-run output before using --apply, and do not log tokens, passwords, private customer data, or raw command output that may contain secrets.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
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 (4)

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The README explicitly describes persistent logging of failures, corrections, patterns, and user preferences, but provides no guidance on excluding secrets, personal data, credentials, or sensitive system details. In a learning/logging skill, these categories can easily capture command output, prompts, tokens, file paths, and behavioral data, creating a privacy and data-retention risk if stored or indexed without minimization or redaction.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The skill persistently stores user corrections, failures, and preferences in a MeiliSearch index and generated files, but the top-level description does not clearly warn users about this retention behavior. That creates a privacy and consent risk because users may reveal sensitive behavioral data or workflow details without understanding they will be logged for future recall.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: When run with --apply, the script automatically extracts content from recent memory notes and sends that content to MeiliSearch using a bearer token, with only a generic flag as user consent. Because the extracted text is heuristic-based and the sensitive-data filter is incomplete, private or security-relevant note content can be exfiltrated to a remote service without meaningful review or confirmation.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The script transmits user-supplied learning content, context, and tags to a remote MeiliSearch service without any inline warning, confirmation, or sanitization policy. In a continuous-learning skill, users may submit sensitive failure details, prompts, tokens, or internal context, so silent exfiltration to a remote index materially increases privacy and data-handling risk.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal