Knowledge Graph - Janusgraph Connector

• Outcome: Review
• Latest audit: Jun 7, 2026 at 4:03 AM UTC
• Version:
  1.0.0

Overview

This JanusGraph skill is mostly coherent, but it presents high-impact database write/delete patterns as production-ready without enough safety controls or warning about unsafe query construction.

Install only if you are comfortable reviewing and constraining its generated Gremlin before execution. Use it against test graphs first, require explicit approval for delete/import/update operations, prefer parameterized bindings over interpolated query strings, and avoid using the provided production-ready examples as-is for user-supplied input.

SkillSpector

By NVIDIA

Vulnerability Patterns

• 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
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (5)

Intent-Code Divergence

Medium

• Confidence: 98% confidence
• Finding: The examples are marketed as 'production-ready' and 'best practices', yet the Python snippets construct Gremlin queries with f-strings and directly embed user-controlled values such as usernames, locations, category names, and titles. This creates a query-injection risk where crafted input can alter traversal logic, access unintended graph data, or potentially trigger destructive operations depending on server permissions and connector behavior.

Intent-Code Divergence

Medium

• Confidence: 97% confidence
• Finding: The summary reinforces that the examples are safe for real-world use while the code repeatedly uses direct dynamic query construction. That mismatch increases the likelihood that developers will copy insecure patterns into deployed systems, making the documentation itself a security hazard rather than a neutral example.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The skill documents destructive Gremlin operations such as dropping vertices and edges without an adjacent warning, confirmation requirement, or guidance on safe scoping. In an agent context, this increases the risk that a user or downstream automation will execute broad deletion commands against a live graph, causing irreversible data loss or service disruption.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The document includes explicit destructive Gremlin patterns such as dropping vertices and edges, but it provides no caution about irreversible data loss, scope validation, or safe-use constraints. In a database connector skill, these examples can be copied directly into production workflows, increasing the likelihood of accidental mass deletion by users or downstream agents.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The import/export section documents graph read/write operations without warning that imports may overwrite or mutate graph state and exports may expose sensitive graph structure or data to files. In the context of a JanusGraph connector skill, this omission is risky because users may treat these as safe operational recipes and unintentionally leak or damage production data.

VirusTotal

VirusTotal findings are pending for this skill version.