Github

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:41 PM UTC
• Version:
  1.0.1

Overview

This is a coherent GitHub integration skill, but it can make real changes to GitHub through a connected Membrane account.

Install only if you trust Membrane and intend to connect GitHub through it. Grant the narrowest GitHub permissions available, review the account or organization being connected, and require explicit approval before allowing the agent to create, update, release, comment, or merge anything.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• 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 (2)

Vague Triggers

Medium

• Confidence: 88% confidence
• Finding: The skill description is overly broad and says to use the skill whenever the user wants to interact with GitHub data, without narrowing scope to safe, intended operations. In an agentic environment, this can cause over-selection of the skill for ambiguous GitHub requests and increase the chance of unintended repository reads or write operations being proposed or executed.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The documentation lists multiple write-capable actions such as creating issues, pull requests, releases, comments, updates, and merges, but does not warn that these operations modify remote user-controlled data and may have irreversible workflow impact. In practice, an agent could treat these as routine actions and perform destructive or sensitive changes without sufficient friction, review, or consent.

VirusTotal

66/66 vendors flagged this skill as clean.