Browser Network Inspector

• Outcome: Review
• Latest audit: May 29, 2026 at 9:29 PM UTC
• Version:
  1.0.1

Overview

This skill appears intended for browser network debugging, but it can capture and save sensitive browser traffic without tight scope or clear user controls.

Install only if you intentionally need browser network debugging. Use it on sites and sessions you are authorized to inspect, prefer an isolated or throwaway browser session, avoid logged-in/private pages unless needed, review generated reports before sharing or committing them, and delete report files when finished.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 89% confidence
• Finding: The skill clearly relies on environment access and browser/network-capable tooling (`agent-browser`, local scripts, file output), but the manifest does not declare permissions. Undeclared capabilities make security review and user consent weaker, and can hide the true operational scope of a skill that captures browser request data and writes reports locally.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The script monkey-patches fetch, XMLHttpRequest, and WebSocket to capture request and response metadata and bodies, then exposes the collected data through global export functions. Even though it attempts to redact common secret fields, it still performs broad interception of potentially sensitive browser traffic without any built-in consent UI, scope confirmation, retention limit, or disclosure mechanism, which creates privacy and misuse risk if invoked unexpectedly or reused in the wrong context.

Missing User Warnings

Medium

• Confidence: 89% confidence
• Finding: The script includes requestBody and responseBody values from captured browser network events directly in the generated Markdown report. Those fields can contain credentials, session tokens, PII, CSRF tokens, API keys, or other secrets from login and API flows, so persisting them to a report creates an unnecessary disclosure risk beyond the original debugging session.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: When an output path is provided, the script writes the full report to disk without any warning that it may contain sensitive network-derived data. Because the report can already include URLs, errors, hosts, and possibly request/response bodies, writing it to a file can leave durable artifacts that are readable later, committed to source control, or collected by backups and telemetry.

VirusTotal

63/63 vendors flagged this skill as clean.