ClawHub
Back to skill

Security audit

Dota2 Coach Publish

Security checks across malware telemetry and agentic risk

Overview

This Dota 2 coaching skill is purpose-aligned; the main risks are broad trigger phrases and manual update scripts that can overwrite local game-data files.

Install only if you want a Dota 2 advice skill that may activate on broad Dota-related wording. Use the bundled data by default, and review/update script paths before running refresh commands, especially the helper scripts with hard-coded OpenClaw workspace paths.

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 (5)

Intent-Code Divergence

Medium
Confidence
97% confidence
Finding
The script prints and records a success message stating the `{s:...}` values were updated when, in fact, it only compares and rewrites text fields in the JSON and never edits the underlying template-value mappings. This can mislead operators into believing a data migration is complete, causing stale or incorrect gameplay data to be published or relied upon without manual follow-up.

Vague Triggers

Medium
Confidence
93% confidence
Finding
The trigger list is broad and includes generic terms like 'dota', 'dota2', and '出装建议', which can overlap with ordinary conversation and cause unintended activation. While this is not a code-execution issue, it can lead to prompt hijacking of normal chats, irrelevant responses, or accidental invocation in contexts where the user did not intend to use the skill.

Vague Triggers

Medium
Confidence
86% confidence
Finding
The skill declares many generic trigger phrases such as "dota", "dota2", and broad help-style phrases, which can cause the skill to activate for loosely related user requests. This is dangerous because it increases the chance of unintended routing or prompt collision with other skills, leading to incorrect task handling or accidental disclosure of skill-specific behavior in contexts where it was not intended.

Missing User Warnings

Medium
Confidence
84% confidence
Finding
The script makes remote network requests and overwrites local JSON database files without an upfront disclosure of those side effects. In an agent-skill context, hidden network access and file writes reduce user consent and auditability, which can lead to unexpected data modification or policy violations even though the destinations are hard-coded and appear legitimate.

External Transmission

Medium
Category
Data Exfiltration
Content
import subprocess

def fetch_ability_details():
    url = "https://api.opendota.com/api/constants/abilities"
    result = subprocess.run(['curl', '-s', url], capture_output=True, text=True)
    return json.loads(result.stdout)
Confidence
77% confidence
Finding
https://api.opendota.com/

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.