a2a-Market-Stake-Freeze
Security checks across malware telemetry and agentic risk
Overview
This is an instruction-only stake-locking policy skill with no executable code, credentials, network access, or hidden behavior in the reviewed artifact.
This skill is reasonable to install as policy-design guidance. Before using it to control real funds or collateral, separately review any runtime implementation, require clear authorization for slashing, validate evidence inputs, keep audit logs, and define rollback or dispute handling.
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
63/63 vendors flagged this skill as clean.