Branerail - CTO-level architectural skill for Claude Code

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:43 PM UTC
• Version:
  1.0.0

Overview

This is a disclosed architecture-review skill with broad activation and imperfect logging examples, but no hidden, destructive, or exfiltrating behavior was found.

Install this if you want an architecture-review skill that may appear frequently during engineering work. Consider narrowing its trigger wording to explicit architecture-review requests, inspect the packaging script before running it, and adapt the logging examples so they do not capture secrets, payment data, or unnecessary personal identifiers.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (7)

Vague Triggers

Medium

• Confidence: 91% confidence
• Finding: The manifest defines very broad trigger keywords such as architecture, design, performance, failure, dependency, and code review, which are common in ordinary developer conversations. In an agent environment, this can cause the skill to activate in contexts far outside its intended scope, potentially overriding more appropriate instructions, increasing prompt surface area, and causing unintended behavior or privilege expansion through excessive guidance injection.

Vague Triggers

Medium

• Confidence: 94% confidence
• Finding: The guide defines very broad trigger keywords such as 'design', 'performance', 'state', 'logging', and 'audit', which are common in normal engineering discussions. In an automatic skill-selection system, this can cause the skill to activate in many unrelated contexts, creating prompt-scope expansion and increasing the chance that architectural instructions override or distract from the user's actual task.

Vague Triggers

Medium

• Confidence: 93% confidence
• Finding: The trigger guidance is unusually broad and includes many common engineering terms such as "design," "performance," "failure," "logging," and "refactor," which can cause the skill to activate in routine conversations where it was not explicitly requested. Unintended activation can bias the agent toward this skill's framing, create prompt-surface expansion, and increase the chance that other embedded instructions or assumptions influence unrelated tasks.

Vague Triggers

Medium

• Confidence: 88% confidence
• Finding: The skill's trigger list is extremely broad and includes many common engineering terms such as 'design', 'performance', 'API', 'optimize', and 'code review'. In an agent environment, this can cause the skill to activate in many unrelated conversations, creating prompt-scope hijacking where the agent's behavior is disproportionately influenced by this skill even when the user did not intend architectural review. The content is not overtly malicious, but the broad activation surface increases the chance of inappropriate instruction injection and workflow disruption.

Vague Triggers

Medium

• Confidence: 93% confidence
• Finding: The skill description explicitly says to use this skill whenever many common architecture- or design-adjacent terms appear and to 'trigger aggressively.' In an agent setting, overly broad activation can cause inappropriate routing, over-collection of context, and unnecessary influence over tasks that do not actually require this skill, increasing the chance of mis-scoped actions or policy bypass through persistent invocation.

Vague Triggers

Medium

• Confidence: 95% confidence
• Finding: The keyword list contains many generic terms like 'design,' 'state,' 'failure,' 'performance,' 'contract,' and 'code review,' combined with instructions to use them liberally. This creates a prompt-injection-like routing hazard where ordinary developer conversations can activate the skill unnecessarily, giving it broad opportunity to shape outputs and consume attention outside its intended scope.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The template explicitly says to log every operation with context and provides an example containing identifiers such as customerId. In a reusable architecture template, this guidance can normalize broad collection of user-linked data in logs without emphasizing data minimization, redaction, or retention controls, increasing the risk of privacy leakage and secondary exposure through log systems.

VirusTotal

65/65 vendors flagged this skill as clean.