ClawHub
Back to skill

Security audit

Koyomi Behavior

Security checks for vulnerabilities and agentic risk

Overview

This is an instruction-only behavior skill that asks the agent to verify work and communicate clearly, with no code, persistence, credentials, or external data flow.

Install this only if you want the agent to apply strict evidence-checking habits broadly. Expect it to run local verification commands and to influence tone/style; avoid it if you need a neutral persona or minimal command use.

SkillSpector

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
94% confidence
Finding
The activation condition is extremely broad: it applies whenever the agent is asked to output a conclusion, complete a task, or make any judgment, which covers a large fraction of normal interactions. Over-broad triggering can unintentionally override user preferences, change response style/behavior across unrelated tasks, and create prompt-scope conflicts that reduce reliability and controllability.

Natural-Language Policy Violations

Medium
Confidence
96% confidence
Finding
The skill hard-codes a Chinese-language/persona mode ('四宫辉夜', '本小姐') without user consent, which can override the user's requested language, tone, or professional style. This is risky because it changes assistant behavior in a way unrelated to task correctness, potentially degrading usability, violating user expectations, and interfering with downstream workflows that require neutral or specific-format output.

Static analysis

No suspicious patterns detected.