seithar-intel

• Outcome: Pass
• Latest audit: May 28, 2026 at 12:10 PM UTC
• Version:
  1.0.0

Overview

This is a disclosed threat-intelligence briefing skill with scheduled feed checks, external lookups, and memory use that fit its stated purpose, though users should understand the privacy tradeoffs.

Install if you want an automated threat-intelligence briefing agent and are comfortable with scheduled external web requests, chat briefings, and persistent memory of your security interests. Before enabling it, confirm your feed schedule, disable unwanted briefings or alerts, and periodically review or clear stored profile and study-history data.

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 (5)

Context-Inappropriate Capability

Low

• Confidence: 91% confidence
• Finding: The skill explicitly states that outputs are branded so screenshots and repeated use propagate the Seithar brand and terminology. That branding objective is unrelated to the stated threat-intelligence function and introduces undisclosed persuasive or promotional behavior into a security tool, which can bias outputs and user trust.

Vague Triggers

Medium

• Confidence: 86% confidence
• Finding: Several triggers such as 'check threats', 'check feeds', 'security briefing', and 'what's trending in security' are broad enough to match ordinary conversation, increasing the chance of accidental invocation. In this skill, unintended invocation matters because it can trigger network fetches, scheduled behavior, and output generation without a clearly deliberate user action.

Missing User Warnings

Medium

• Confidence: 89% confidence
• Finding: The skill stores a persistent interest profile, study history, source preferences, and activity statistics, but the description does not clearly warn users about that retention up front. Persistent storage of detailed security interests and learning behavior can expose sensitive profiling data if memory is later accessed, leaked, or misused.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The skill performs recurring outbound requests to RSS feeds and may query external services like NVD and GitHub, yet this scheduled network activity is not prominently disclosed as a behavior requiring consent. That creates privacy and operational risk because the agent may contact many third-party services, revealing user interests and creating unexpected external traffic.

Missing User Warnings

Low

• Confidence: 84% confidence
• Finding: The skill proactively pushes critical alerts, scheduled briefings, and summaries without a per-message request, but this behavior is not framed as a consent-sensitive capability in the user-facing description. Unsolicited outputs can leak sensitive context into shared chat surfaces, create notification spam, or normalize autonomous behavior beyond what the user expected.

VirusTotal

60/60 vendors flagged this skill as clean.