Security audit

Group Debug & Deploy Expert

Security checks across malware telemetry and agentic risk

Overview

This skill is a powerful DevOps assistant with coherent goals, but it asks for broad command authority and persistent project memory in ways users should review before installing.

Install only if you want a high-authority DevOps skill and can supervise commands before they run. Remove or parameterize the bundled BeijiXing-AI project context, do not let repository-local rule files override your agent or platform rules, define which credentials and production systems are in scope, and require redaction or deletion controls for checkpoints, logs, and archives.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Vague Triggers

High

Confidence: 92% confidence
Finding: The invocation scope is so broad that the skill can be activated for routine development, debugging, deployment, and operations tasks across many contexts. In a skill that also requests terminal execution, file writes, and process spawning, overly broad triggers increase the chance of unnecessary high-privilege activation and unintended system-impacting actions.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill advertises powerful capabilities including file.write, terminal.execute, and process.spawn without an upfront concise danger notice or clear consent gate. That combination can normalize risky autonomous behavior and makes accidental destructive use much more likely, especially when the skill claims broad applicability.

Ssd 3

Medium

Confidence: 96% confidence
Finding: The skill mandates persistent logging and archival of task inputs, outputs, evidence, and session state, which is likely to capture secrets, credentials, personal data, logs, and infrastructure details during normal debugging. Persistent natural-language records materially increase the blast radius of any later compromise or accidental disclosure.

Ssd 3

Medium

Confidence: 97% confidence
Finding: The Problem Resolution Archive explicitly requires raw evidence and full command output, which commonly contain access tokens, headers, stack traces, hostnames, credentials, and customer data. A mandatory archive of verbatim operational evidence creates a durable sensitive-data repository that attackers or other projects could mine later.

Ssd 3

Medium

Confidence: 98% confidence
Finding: The checkpoint/resume/handoff system instructs the agent to persist full workflow state across sessions, including verified facts, pending issues, changed files, and critical context. This creates a semantic memory layer that can preserve sensitive project state and makes cross-session or cross-project leakage more likely if isolation or access controls fail even once.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.