ClawHub
Back to skill

Security audit

企业碳资信评级报告

Security checks across malware telemetry and agentic risk

Overview

This skill coherently reads user-provided business materials, confirms extracted rating data, calculates a carbon credit score, and creates a Word report without hidden network, credential, persistence, or destructive behavior.

Install this skill when you intend to process enterprise carbon-rating materials locally and generate a Word report. Review the extracted fields and missing-data table before approving scoring, and treat both the input materials and generated DOCX/JSON outputs as sensitive business records.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Lp3

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding
The skill explicitly instructs the agent to read customer files and generate output artifacts, and it invokes local scripts that imply file read/write behavior, yet no permissions are declared. This creates a capability/authorization mismatch: an operator may not realize the skill can access input documents and create report files, weakening review, isolation, and consent controls around sensitive enterprise data.

Missing User Warnings

Low
Confidence
83% confidence
Finding
The skill processes customer-provided corporate materials and produces a formal Word report, but it does not warn about sensitive data handling, artifact creation, or downstream sharing/storage risks. In a workflow involving business identity data, compliance records, and financial/environmental indicators, this omission can lead to unintended disclosure or retention of sensitive information.

Natural-Language Policy Violations

Medium
Confidence
89% confidence
Finding
The skill metadata hard-codes Chinese-language interaction and output in the display name, descriptions, and default prompt without any indication that language selection is based on user preference or locale. This can cause user misunderstanding, inaccessible output, and unsafe downstream use if users cannot correctly review extracted fields, missing-data prompts, or generated reports in a language they understand.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.