ClawHub
Back to skill

Security audit

Smart Procurement Navigator 2.0.0

Security checks across malware telemetry and agentic risk

Overview

The skill is mostly a disclosed procurement assistant, but it includes procurement-intelligence prompts that could encourage collecting non-public influence information about evaluators or decision-makers.

Review before installing in real procurement work. Use it only with authorized documents, disable or rewrite prompts that ask for evaluator relationships, internal decision processes, or non-public key-person intelligence, and keep final bid decisions, pricing, commitments, signatures, and submissions under human control.

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 (3)

Context-Inappropriate Capability

Medium
Confidence
94% confidence
Finding
The template explicitly asks users to build a relationship map involving the procuring entity, agency, and evaluation experts. In a procurement workflow, collecting or inferring ties to评标专家 can facilitate improper influence, conflict exploitation, or bid-rigging behavior rather than legitimate document analysis. The skill context makes this more dangerous because it is embedded in a bid/no-bid decision aid, where such intelligence could be operationalized to manipulate the process.

Context-Inappropriate Capability

Medium
Confidence
92% confidence
Finding
Requesting the procuring entity's internal decision process and key-person information goes beyond normal procurement document support and encourages collection of non-public influence intelligence. That information could be used to target decision-makers, bypass formal channels, or gain unfair competitive advantage. In this procurement-assistance context, the request is especially sensitive because it directly supports strategy around a live tender.

Vague Triggers

Medium
Confidence
94% confidence
Finding
The README advertises activation via very generic natural-language requests such as analyzing procurement notices or generating response structures, without defining strong scope boundaries, required confirmations, or safe input constraints. In an agent-skill context, broad triggers increase the chance of unintended activation on unrelated user content or mixed-trust documents, which can cause overreach into sensitive procurement, legal, or compliance workflows.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.