ClawHub
Back to skill

Security audit

Agentvault

Security checks across malware telemetry and agentic risk

Overview

AgentVault is a coherent local credential-vault skill, but it needs review because it exposes secret retrieval, MCP signing, wallet-key handling, and plaintext export paths without fully clear approval boundaries.

Install only if you trust the external npm package and need an agent-facing credential vault. Set a personal rule that the agent must ask before retrieving any secret value, exporting vault data, using plaintext export, starting MCP, signing anything, or touching wallet/private-key commands. Connect MCP only to trusted local clients and prefer encrypted backups over decrypted JSON.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Description-Behavior Mismatch

Medium
Confidence
89% confidence
Finding
The documented wallet commands extend the skill from credential vaulting into cryptocurrency private-key generation and import, which is outside the stated scope of secret sandboxing and encrypted memory. That scope expansion increases the chance that users or downstream agents will handle highly sensitive signing keys without corresponding security controls, review, or justification.

Context-Inappropriate Capability

Medium
Confidence
92% confidence
Finding
Exposing an MCP signing tool such as `vault.sign_x402` gives agents the ability to perform cryptographic signing, which can authorize transactions or authenticated actions far beyond simple secret storage. In an agent-facing interface, signing is especially dangerous because it can turn read access to secrets into active misuse of credentials or financial authority.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The command reference allows exporting the vault in decrypted plaintext JSON and only notes a flag requirement, not the security consequences of producing an unencrypted secrets dump. This is dangerous because users or agents may create a plaintext archive that can be copied, logged, indexed, or exfiltrated much more easily than the encrypted vault.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.