ClawHub
Back to skill

Security audit

Latchshot

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed Latchshot connector, but users should know that page capture creates an uploaded artifact.

Install only if you intend to use OOMOL's `oo` CLI with a connected Latchshot account. Treat page capture as creating and uploading a rendered artifact, and confirm the target URL before capturing pages that may expose private or sensitive content.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Description-Behavior Mismatch

Medium
Confidence
96% confidence
Finding
The manifest says the skill is for 'searching and reading data,' but the documented `capture_page` action performs a side effect by rendering arbitrary public URLs and uploading the resulting artifact to transit storage. That mismatch can cause downstream agents or users to treat the skill as read-only and invoke it in contexts where artifact creation, storage, or retrieval of sensitive page content was not intended.

Intent-Code Divergence

Medium
Confidence
98% confidence
Finding
The safety section states that untagged actions are reads and safe to run directly, yet `capture_page` is untagged even though it creates a rendered artifact and uploads it. This incorrect safety labeling may lead an agent to execute the action without confirmation, enabling unintended external requests, artifact generation, and persistence of captured content.

Vague Triggers

Medium
Confidence
90% confidence
Finding
The instruction to use this skill for 'ANY Latchshot request' is overly broad and encourages automatic routing without evaluating whether the requested operation is appropriate, necessary, or safely scoped. In combination with the misleading read-only framing, this increases the chance that page-capture actions are triggered in situations where a narrower, more explicit activation rule should have applied.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.