External Ki Integration Backup

• Outcome: Review
• Latest audit: May 28, 2026, 6:37 AM
• Version:
  1.0.0

Overview

This skill has a coherent purpose, but it can send user content and credentials-backed requests to third-party AI services with broad activation and insufficient per-use consent controls.

Install only if you want your agent to consult third-party AI services. Before each use, confirm the target provider, what content will be sent, whether costs may be incurred, and whether any logs or memory entries will be written. Avoid confidential, regulated, personal, financial, or secret-bearing content unless you have explicitly approved that provider and data flow.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• 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

Findings (4)

Context-Inappropriate Capability

Medium

• Confidence: 93% confidence
• Finding: The skill directs logging external-AI costs and interactions to local files/memory without establishing a strict need, retention limit, or consent boundary. Because prompts sent to external models may contain sensitive user task content, persisting metadata or interaction details can create unnecessary secondary data retention and increase privacy and compliance risk.

Context-Inappropriate Capability

Medium

• Confidence: 96% confidence
• Finding: The example workflow explicitly instructs the agent to log the interaction in memory, extending retention beyond the core purpose of obtaining an external model response. This can preserve user prompts, generated outputs, or behavioral patterns that were only meant for transient task execution, increasing the blast radius of later memory disclosure or misuse.

Vague Triggers

Medium

• Confidence: 89% confidence
• Finding: The skill's activation criteria are broad and normalize consulting third-party AI services for many task types without strong exclusions for sensitive, regulated, or user-local-only content. In practice this can cause unnecessary external disclosure of user data or overuse of external services when internal handling would be safer.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill describes sending prompts and task data to external AI systems but does not require a clear user-facing warning and explicit confirmation before transmission. Since users may not realize that their content will leave the local environment and be processed under third-party policies, this creates a meaningful privacy and consent failure.

VirusTotal

58/58 vendors flagged this skill as clean.