ClawHub
Back to skill

Security audit

Linkup

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only Linkup search guide, but it gives broad scraping guidance and explicitly supports collecting LinkedIn comments and commenter profile URLs without privacy guardrails.

Install only if you intend agents to use Linkup for web retrieval. Keep it constrained to user-authorized, task-relevant sources; avoid bulk LinkedIn comment or commenter-profile collection unless there is a legitimate compliant need, and prefer narrower standard searches when deep scraping is unnecessary.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (4)

Context-Inappropriate Capability

Medium
Confidence
90% confidence
Finding
The skill explicitly teaches extraction of LinkedIn profile details, posts, comments, and commenter profile URLs, which expands from generic web search guidance into social-data harvesting of third parties. That creates privacy and abuse risk because an agent can be directed to collect and aggregate personal or behavioral data at scale without any consent, necessity check, or policy guardrails.

Vague Triggers

Medium
Confidence
83% confidence
Finding
The activation scope is so broad that it can trigger on many ordinary research tasks, increasing the chance this skill is invoked when unnecessary. In context, that matters because the skill encourages deep retrieval and scraping behaviors, so over-triggering can cause needless external data collection and expansion of task scope beyond user intent.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The LinkedIn scraping instructions omit any warning about privacy, personal-data handling, or platform-policy implications despite enabling extraction of profile details, posts, comments, and commenter identities. Without guardrails, an agent may treat social-data collection as routine and return sensitive third-party information in ways that users did not explicitly request.

Ssd 3

Medium
Confidence
93% confidence
Finding
These instructions directly support bulk collection of third-party LinkedIn comments and commenter profile URLs in plain language, making misuse straightforward. In the context of a general-purpose search skill, this materially increases the risk of social graph harvesting, profiling, and surveillance-like aggregation of individuals unrelated to the primary task.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.