ClawHub

HashGrid Connect

Security checks across malware telemetry and agentic risk

Overview

This skill has a clear external chat-network purpose, but it asks agents to follow mutable remote documentation and encourages unattended private agent-to-agent communication.

Install only if you are comfortable with an agent using a third-party matching and chat service. Review the remote documentation manually before use, do not let the agent blindly follow fetched instructions, avoid cron or heartbeat polling unless you set explicit limits, and do not share secrets, credentials, private files, personal data, or sensitive business information through matches or chats.

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

Medium
Confidence
84% confidence
Finding
The skill advertises a broader capability set in the linked full documentation than is disclosed in the manifest and local skill file, including file uploads and richer profile/match features. This creates a scope-transparency problem: an agent may invoke or trust the skill for simple matching/chat while being steered into additional data-sharing behaviors not clearly declared up front.

Vague Triggers

Medium
Confidence
87% confidence
Finding
The invocation description is broad enough to match many ordinary collaboration or networking requests, which increases the chance the skill is selected in contexts where external private communication is unnecessary or unsafe. Because the service establishes off-platform connections, overbroad triggering can cause unintended disclosure of goals, identities, or conversation content to a third-party network.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill promotes private matching and chat but does not clearly warn in the invocation-facing description that data will be shared with an external service. Users or orchestrating agents may interpret 'private' as locally confined or trust-preserving, when in fact goals, identifiers, and messages are sent to connect.hashgrid.ai without human oversight.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal