New Relic

• Outcome: Review
• Latest audit: May 29, 2026, 9:38 PM
• Version:
  1.0.4

Overview

This appears to be a real New Relic integration, but it gives an agent broad account-changing and raw API access without clear confirmation safeguards.

Install only if you trust Membrane and intend to grant it New Relic access. Use a least-privilege New Relic account, review the Membrane connection scopes, prefer listed read-only actions, and require explicit confirmation with the exact resource ID before create, update, delete, or proxy requests.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• 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

Findings (4)

Description-Behavior Mismatch

Medium

• Confidence: 85% confidence
• Finding: The skill is branded and described as a New Relic integration, but its connection flow explicitly allows Membrane to create a new app/connector automatically when the URL does not match a known app. That expands the operational scope beyond the declared integration and could let an agent pivot into arbitrary connector creation or unintended third-party access paths if the input URL is influenced by user prompts or model error.

Description-Behavior Mismatch

Medium

• Confidence: 92% confidence
• Finding: The documented proxy request capability permits direct requests to arbitrary API paths through the authenticated Membrane connection, which is broader than the listed New Relic account/data actions. This increases the chance of unauthorized or overly powerful operations, including invoking undocumented, destructive, or higher-risk endpoints without the guardrails of typed actions.

Vague Triggers

Medium

• Confidence: 81% confidence
• Finding: The description 'Use when the user wants to interact with New Relic data' is broad and could trigger the skill for loosely related prompts, causing the agent to invoke account-connected actions when the user only wanted general information or analysis. In a capability-bearing integration, over-broad activation increases the risk of unintended external calls and side effects.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill advertises a delete-application action without any warning, confirmation, or safety guidance around destructive operations. In an agent context, this omission makes accidental or prompt-induced deletion more likely, especially if the model interprets ambiguous user requests as authorization to modify or remove production resources.

VirusTotal

63/63 vendors flagged this skill as clean.