Nonprofit RBM Logic Model
Security checks across malware telemetry and agentic risk
Overview
This is an instruction-only nonprofit grant and logic-model drafting skill with no code, credentials, or hidden system access.
Reasonable to install as a drafting and review aid. Users should still verify donor requirements, evidence, budgets, legal/compliance issues, and any sensitive proposal content before relying on the output for submission.
SkillSpector
By NVIDIA
Vulnerability Patterns
- 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
- Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
VirusTotal
64/64 vendors flagged this skill as clean.