Longrunning Agent

Security checks across malware telemetry and agentic risk

Overview

This is a mostly transparent project-tracking workflow skill, with expected cautions around local setup scripts, commits, and possible web logging.

Install only if you are comfortable with an agent maintaining project task files and making commits. Review any init.sh before it runs, inspect changes before accepting commits, avoid putting secrets in progress or task logs, and confirm how the referenced web UI stores or syncs session output before using it on private work.

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 (1)

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The skill states that it integrates with a web app and that tasks, progress entries, session output, and git commits are logged or synced, but the description and earlier workflow guidance do not clearly warn users about this external data flow. This creates a real transparency and privacy issue because agents may process sensitive code, secrets, internal project details, or commit metadata without the operator realizing that the information may be transmitted or retained outside the local environment.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal