Rubicon Sentinel v2 — Sovereign Forge

• Outcome: Review
• Latest audit: May 28, 2026, 11:36 AM
• Version:
  2.0.0

Overview

This no-code political analysis skill is not malware, but its scoring rules include identity-based and nativist criteria that can drive roasts, memes, and tweet drafts, so it should be reviewed before installation.

Install only if you explicitly want a strongly opinionated sovereignty-analysis skill. Treat its scores as ideological commentary rather than neutral intelligence, verify cited sources before relying on outputs, avoid private scan topics, and do not expose a Twitter bearer token unless you intentionally want X/Twitter sentiment lookup.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• 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 (6)

Vague Triggers

Medium

• Confidence: 86% confidence
• Finding: The trigger phrase at line 23 is broad enough that normal conversation containing similar wording could unintentionally activate the skill and cause external searches or formatted political analysis without clear user intent. Because this skill automatically performs web_search and may chain into other modes, ambiguous invocation increases the chance of unintended tool use and surprising behavior.

Vague Triggers

Medium

• Confidence: 83% confidence
• Finding: Using 'etc.' in the trigger list leaves the activation surface undefined, making it impossible for users and reviewers to know what phrases may invoke the skill. In practice this can lead to over-broad matching, accidental activation, and unexpected execution of network-backed actions.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The skill can query X/Twitter using a bearer token, but that capability is not clearly disclosed in the top-level description where users decide whether to invoke it. Hidden or under-disclosed third-party access is risky because user prompts may cause external data sharing, and token-backed requests can have privacy, compliance, and cost implications.

Natural-Language Policy Violations

High

• Confidence: 98% confidence
• Finding: This rubric hard-codes exclusionary political claims such as 'mass uncontrolled migration,' 'elite self-hatred,' and 'low crime in core populations' as if they were objective security metrics. In this skill’s context, those criteria are used to drive outputs, scores, and recommendations, which can systematically launder ideological bias into seemingly factual analysis and produce discriminatory or extremist content.

Natural-Language Policy Violations

High

• Confidence: 99% confidence
• Finding: The demographic scoring explicitly privileges 'native stock' and labels some migration as 'incompatible,' which is nativist framing rather than a legitimate security measure. Because the skill is designed to generate scans, forecasts, roasts, and persuasive outputs, this can operationalize discriminatory ideology at scale and normalize targeting of protected or national-origin groups.

Natural-Language Policy Violations

Medium

• Confidence: 96% confidence
• Finding: This section mandates a preferred 'civilizational' worldview and rewards 'unapologetic celebration of Western achievements,' turning the tool into an ideological enforcement mechanism rather than an analytical rubric. In the broader skill context—marketed as delivering 'truth over comfort,' 'red pills,' and 'roasts'—this increases the risk of propagandistic, polarizing, and potentially hateful outputs under the guise of objective scoring.

VirusTotal

65/65 vendors flagged this skill as clean.