God's eye view of your dev repos. Multi-project tracking across GitHub/Azure DevOps. AI learns from your commits to upgrade your agents.md.

• Outcome: Pass
• Latest audit: May 28, 2026, 1:54 AM
• Version:
  0.1.0

Overview

This skill is a disclosed repo-dashboard and agent-coaching tool that uses your GitHub CLI access and stores project context locally.

Install only if you are comfortable with this tool reading configured repositories through your existing gh login and caching repo activity plus agent-instruction content under ~/.god-mode. Avoid running agent analysis in CI or shared terminals unless you are comfortable exposing the generated prompt output.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• 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 (3)

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The invocation text is broad enough that an orchestrating agent may call this skill in many repo-analysis or project-status contexts, even when the user did not intend network sync, config edits, or persistent caching. Over-broad activation increases the chance of unnecessary data access and side effects across multiple repositories.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The command prints a fully assembled LLM prompt containing raw agent instructions and recent commit messages directly to stdout. Those contents may include sensitive operational guidance, internal repository details, secrets accidentally committed to instructions, or confidential commit text, and there is no consent gate, redaction, or warning before disclosure. In a developer tooling context that aggregates repo state, this increases the chance of unintended exposure through terminal logs, CI logs, shell history capture, or copy/paste into external LLM systems.

Missing User Warnings

Medium

• Confidence: 85% confidence
• Finding: The commit-fetching helper collects and returns author email addresses from commit metadata, which can expose personal or internal corporate email addresses to downstream agent components that may not need them. In a cross-repo oversight skill, this increases privacy and data-minimization risk because the function can aggregate contributor identities at scale without any access scoping or field minimization.

VirusTotal

54/54 vendors flagged this skill as clean.