商业数据洞察 / Biz Data Insight

• Outcome: Review
• Latest audit: May 29, 2026 · 9:26 PM UTC
• Version: 1.0.0

Overview

This appears to be a legitimate business analytics skill, but it needs review because setup can expose raw rows from configured databases or files before users may expect it.

Install only if you are comfortable giving the skill read access to the selected datasource. Use read-only, least-privilege database credentials or sanitized reporting views, avoid broad production databases with PII or financial records, and review SQL before confirming execution.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration

Findings (7)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 93% confidence
• Finding: The skill invokes local scripts, reads environment variables including credentials, accepts file paths/URLs, and connects to external data sources, yet it declares no explicit permissions. This creates a capability/permission mismatch that can cause the host to grant more access than users or policy expect, increasing the risk of unintended network access, file access, or secret use during execution.

Description-Behavior Mismatch

Medium

• Confidence: 94% confidence
• Finding: The schema exploration routine goes beyond metadata discovery and returns up to 5 rows of actual table/file contents. In a business-data connector, those samples may contain PII, financial records, secrets, or regulated data, so the feature creates an unnecessary data exposure channel during a supposedly low-risk 'explore schema' action.

Vague Triggers

Medium

• Confidence: 85% confidence
• Finding: Using trigger language such as 'or similar intent' for datasource configuration makes activation boundaries ambiguous. An orchestrator or agent may invoke the skill for loosely related messages, causing premature collection of connection details or attempts to access configured data sources without clear user intent.

Vague Triggers

Medium

• Confidence: 87% confidence
• Finding: The trigger condition 'data analysis type questions' is overly broad for a skill that can generate SQL and query live business data. Overbroad activation increases the chance that ordinary conversation or high-level brainstorming gets turned into actual database interaction, exposing sensitive data or consuming quotas unexpectedly.

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The report-generation flow is triggered by 'or similar intent' despite being capable of running multiple queries and generating derived outputs. This ambiguity can lead to unintended scheduled/report-like processing from casual requests, increasing the chance of unnecessary data access and disclosure in generated reports.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The database exploration path returns sample table contents without clear disclosure or consent, which can surprise users who expected only schema inspection. Because this tool targets business datasets, even a five-row sample may reveal sensitive customer, payment, or operational information and can materially expand data exposure.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The file exploration path serializes and outputs the first five records from the provided file, again exceeding pure schema discovery. For local CSV/Excel/JSON inputs, this may disclose private business records or regulated data through command output or downstream logs without the user's informed consent.

VirusTotal

64/64 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.