ClawHub

xAPI

Security checks across malware telemetry and agentic risk

Overview

This looks like a legitimate xapi CLI skill, but it gives agents broad external API, social posting, OAuth, and payment-related authority without strong consent boundaries.

Install only if you trust xapi.to and the xapi-to npm package. Treat this as a broad external API gateway, not a read-only lookup tool. Require explicit approval before OAuth binding, POST/DELETE/write calls, Twitter posting, account registration, or top-ups, and avoid sending secrets or sensitive documents through it unless external processing is acceptable.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
94% confidence
Finding
The skill metadata presents xapi mainly as a data lookup and text-processing tool, but the body also enables materially different side-effecting operations: posting tweets, binding OAuth accounts, registering accounts, and topping up funds. This mismatch can cause an agent or user to invoke write, auth, or financial actions without adequate expectation-setting or consent boundaries, increasing the risk of unauthorized external changes or charges.

Context-Inappropriate Capability

High
Confidence
97% confidence
Finding
Including `topup` payment operations in a skill advertised for external data access introduces direct financial side effects unrelated to the primary lookup use case. If an agent follows the documented remediation flow for insufficient balance, it could trigger purchases or expose payment URLs containing sensitive tokens, leading to monetary loss or credential leakage.

Vague Triggers

High
Confidence
91% confidence
Finding
The trigger text is extremely broad, including requests to search the web, news, summarize or rewrite text, generate embeddings, or 'call any third-party API through xapi,' plus activating when the user merely mentions xapi. Overbroad activation increases the chance that the skill is selected for many unrelated requests, causing unnecessary data transmission to third-party services and exposing users to hidden external side effects.

Missing User Warnings

Medium
Confidence
89% confidence
Finding
The skill encourages networked calls to multiple third-party services, OAuth binding, account registration, and payment actions, but does not foreground user-facing warnings about what data is sent externally or which actions have persistent or financial side effects. That makes accidental disclosure of prompts, personal data, account linkage, or unintended transactions more likely in normal use.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal