Escalate

Security checks across malware telemetry and agentic risk

Overview

Escalate is a disclosed local memory skill for deciding when an agent should ask before acting, with clear approval gates for risky actions.

Install this if you want the agent to remember approval preferences across sessions. Review ~/escalate/ periodically, keep secrets out of it, and only approve proposed AGENTS or SOUL snippets after reading the exact text.

SkillSpector

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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)

Vague Triggers

Medium

Confidence: 87% confidence
Finding: This markdown file is in scope for vague-trigger review. The instruction to save "activation rules for when the skill should step in" is high-level but does not define specific invocation phrases, boundaries, or negative examples, which could lead to overly broad or inconsistent activation in normal conversation.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal