KWDB Intelligent Inspection

• Outcome: Review
• Latest audit: Jul 23, 2026 · 6:33 AM UTC
• Version: 1.2.0

Overview

This KaiwuDB inspection skill is purposeful, but it should be reviewed because it can collect and display sensitive database telemetry over unencrypted HTTP.

Install only if you are comfortable letting the agent probe KaiwuDB hosts and collect database metrics, including slow SQL statement text and error details. Use it only against systems you administer, prefer localhost, VPN, or SSH-tunneled access, and treat generated reports and script output as confidential operational data.

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
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 88% confidence
• Finding: The skill instructs the agent to run local scripts under `scripts/` and perform connectivity probes, which implies shell or code-execution capability, yet no corresponding permissions are declared. This creates a mismatch between documented behavior and declared access, increasing the risk of unintended or insufficiently governed command execution in the agent runtime.

Missing User Warnings

Medium

• Confidence: 89% confidence
• Finding: The documentation explicitly says the script outputs SQL statement text, application names, database users, database names, and error messages, but it does not warn that these fields can contain sensitive data such as secrets embedded in queries, tenant identifiers, schema details, or operationally sensitive errors. In the context of a database inspection skill, this increases the chance operators will run the tool broadly and share outputs without redaction, causing unintended disclosure of sensitive telemetry.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The documentation states the script uses the KaiwuDB Statements API over plain HTTP and notes TLS is unsupported, but it does not warn that this exposes database telemetry in cleartext to interception or modification on the network. Because this skill is specifically for health checks and inspection, users may run it against production clusters where statement data and metadata are especially sensitive, making the omission more dangerous.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The script retrieves and prints raw statement telemetry including query text, usernames, database names, and last error strings. In a database inspection skill, this is especially sensitive because SQL text and errors often contain secrets, schema details, tenant identifiers, or personal data, and the script emits them directly to stdout without warning, masking, or access control considerations.

Overly Broad Trigger

Low

• Category: Trigger Abuse
• Confidence: 81% confidence
• Finding: The trigger `巡检` is extremely broad and can match many unrelated user requests involving generic inspection concepts, causing accidental invocation of a skill that performs network probing and script execution. In a security-sensitive operational skill, overbroad activation increases the chance of unintended data collection or infrastructure interaction.

VirusTotal

VirusTotal findings are pending for this skill version.

Static analysis

No suspicious patterns detected.