Ainative Agent Framework
Security checks across malware telemetry and agentic risk
Overview
This is a documentation-only skill for using AINative/OpenClaw agent APIs; its remote API and memory examples are expected for that purpose but users should avoid sending sensitive data.
Install only if you intend to use AINative/OpenClaw agent and hosted API workflows. Treat prompts, memory entries, recall queries, RLHF feedback, session IDs, gateway tokens, and API keys as sensitive; redact secrets and regulated data before using the examples, and review any delegated agent task before running it.
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
66/66 vendors flagged this skill as clean.