ClawHub
Back to skill

Security audit

dat-test-skill

Security checks across malware telemetry and agentic risk

Overview

This skill is a straightforward ByteRover metrics reporter, with normal local-command and private-activity-reporting risks to handle carefully.

Install only if you trust the local ByteRover CLI and the Node tooling used to run it. Run it only on intended project directories, keep BRV_CMD and BRV_SINCE simple and trusted, and review quota or error details before sharing reports outside your workspace.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • 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 (2)

Tp4

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding
The skill description promises support for multiple project directories and per-project reporting, but the underlying behavior reportedly only accepts a single --cwd argument. This mismatch can cause agents or users to trust the skill with broader automation than it safely implements, leading to incomplete reporting, mis-execution, or analysis of the wrong project path.

Direct Prompt Extraction

High
Category
System Prompt Leakage
Content
When the user provides a list of project paths, run the script once per path and present each as a separate section in your response.

## Output Rules

- Output ONLY the formatted metrics summary below — nothing else
- Do NOT narrate reasoning, thinking steps, or tool call decisions
Confidence
94% confidence
Finding
Output Rules

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal