Harness Evolve

Security checks across malware telemetry and agentic risk

Overview

This skill is transparent about self-checking and maintenance, but it can automatically change and commit project files and may run a configured shell notification command.

Install only if you want an agent to perform project maintenance, not just analysis. Define safe_files and config_files explicitly, run it on a branch, review diffs and commits, keep runtime agent configuration proposal-only, and do not configure notify_command unless the command and repository configuration are fully trusted.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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 (1)

Context-Inappropriate Capability

Medium

Confidence: 95% confidence
Finding: The skill allows execution of an arbitrary user-configured shell command via notify_command. Even though it is framed as optional notification behavior, this expands the skill from self-check/proposal generation into command execution, creating a command-injection and unintended side-effect surface if the template or substituted message is unsafe.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal