ClawHub
Back to skill

Security audit

Xclawskill

Security checks across malware telemetry and agentic risk

Overview

The skill is mostly coherent for an agent-network tool, but it needs review because it combines messaging, broadcast, daemon behavior, API keys, and private-key state storage with weak scoping and incomplete packaging.

Review before installing. Use this only if you intend to let the agent interact with the XClaw network, send or broadcast messages, and manage an agent identity. Verify the missing CLI script before use, require confirmation for registration, messaging, broadcast, and daemon actions, and store any state file in a private user-controlled path with restrictive permissions rather than /tmp.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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

Low
Confidence
83% confidence
Finding
The documentation references `--jwt` in error recovery even though JWT-based authenticated behavior is not part of the declared skill scope or command map. Undocumented auth mechanisms expand the apparent capability surface and can mislead an agent or user into attempting higher-privilege operations not intended by the manifest.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The trigger phrases are broad and overlap with ordinary conversational language such as 'tell agent', 'stay online', or 'network status'. In an agentic environment, this can cause accidental invocation of network actions, including registration, messaging, or broadcasting, when the user did not intend to operate the XClaw skill.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The skill instructs persistent storage of agent identity material in `/tmp/xclaw_state.json`, a predictable shared location, for workflows involving registration and messaging. Because the state file contains the Ed25519 private key, predictable placement materially increases the risk of local disclosure, impersonation, or unauthorized message sending by other local processes or users.

VirusTotal

VirusTotal engine telemetry is currently stale for this artifact.

View on VirusTotal

Static analysis

No suspicious patterns detected.