ClawHub
Back to skill

Security audit

dawn-code-master

Security checks for vulnerabilities and agentic risk

Overview

This is a broad coding-workflow skill, but its behavior is disclosed, purpose-aligned, and limited to markdown instructions rather than hidden code or persistence.

Install this only if you want a broad, opinionated coding workflow to guide many development tasks. Expect it to push specs, tests, reviews, and gated workflows; it does not appear to add hidden execution or persistence.

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 (3)

Vague Triggers

Medium
Confidence
88% confidence
Finding
The skill description is extremely broad ('write code, fix bugs, review PRs, do design') and overlaps with ordinary coding assistance requests, which makes unintentional activation likely. In an agentic environment with exec and workspace write permissions, over-broad triggering can cause the skill to steer behavior or unlock workflows in contexts the user did not explicitly request.

Vague Triggers

Medium
Confidence
84% confidence
Finding
The routing table maps vague conditions like 'implementing code,' 'UI work,' or 'security concerns' to internal modes, which can cause the skill to self-activate across a very wide range of normal interactions. This is risky because the skill is effectively positioning itself as a default controller for many coding tasks, increasing prompt-scope capture and reducing user control over when the behavior applies.

Natural-Language Policy Violations

High
Confidence
92% confidence
Finding
The description states '写代码、修bug、审PR、做设计，按这个来,' which directs interactions toward Chinese-language operation and workflow control without explicit user consent. Forcing output language or interaction style can override user preferences, degrade comprehension, and create a prompt-injection-like control channel when bundled into a broad, auto-invoked skill.

Static analysis

No suspicious patterns detected.