Langgraph Implementation
Security checks across malware telemetry and agentic risk
Overview
This is a documentation-only LangGraph guide with expected persistence and multi-agent examples, and no evidence of hidden code or credential use.
Safe to install as a reference skill. When copying its patterns into production, decide what conversation state may be saved, how long checkpoints are retained, and which agents or tools may receive user messages.
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
65/65 vendors flagged this skill as clean.