skill-coach

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:34 PM UTC
• Version:
  1.0.0

Overview

This is an instruction-only skill-coaching package whose behavior matches its stated purpose, with a minor risk that its trigger examples could be too broad.

Install is reasonable for a skill-authoring workflow. Review generated skills before enabling or publishing them, make trigger phrases specific to the intended domain, and avoid adding scripts, API access, or credential handling to generated skills unless the purpose and user controls are explicit.

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 (3)

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: Broad trigger phrases like 'create skill' and 'build skill' can overlap with normal user requests, causing unintended activation of this coaching skill when the user meant something else. In an agent ecosystem, accidental invocation can redirect the conversation, collect unintended input, or interfere with more appropriate skills, increasing prompt-routing and authorization risk.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The skill explicitly recommends broad trigger phrases such as "analyze data" and generic noun phrases, which can match many ordinary user requests outside the intended context. In an agent/skill system, ambiguous activation boundaries can cause accidental invocation or routing to this skill when a user did not intend to enter a skill-creation workflow, creating prompt-scope confusion and increasing the chance of misuse or interference with other skills.

Vague Triggers

Medium

• Confidence: 86% confidence
• Finding: The quick-start template asks only for three trigger keywords and does not require specificity, namespace, or activation constraints, leaving invocation boundaries underspecified. That makes it easy for generated skills to inherit overly broad triggers, which can lead to unintended activation, collisions with other skills, and confused delegation behavior across the agent environment.

VirusTotal

63/63 vendors flagged this skill as clean.