ClawHub
Back to skill

Security audit

知识体系建构指南

Security checks for vulnerabilities and agentic risk

Overview

This is a documentation-only knowledge-organization guide with a disclosed local write-back workflow that users should keep under review.

Install this if you want a guide for structuring a personal knowledge base. Before allowing any write-back, define the exact raw/ and wiki/ folders the agent may read or modify, preview generated notes, and keep human review so AI-generated summaries do not become trusted notes without verification.

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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
88% confidence
Finding
The trigger section uses broad, natural phrases such as asking why learned knowledge is unusable or asking for analysis/help organizing knowledge. These phrases are common in ordinary conversation, so the skill may activate unintentionally and steer the agent into a specialized workflow the user did not explicitly request, including knowledge-base actions described elsewhere in the file.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill instructs writing synthesized answers back into a wiki/knowledge base and emphasizes '回写' as a key action, but it does not clearly warn that this modifies local files or require explicit user confirmation. In an agent environment with filesystem access, this can cause unintended persistence, overwrite trusted notes, or introduce inaccurate AI-generated content into a user's knowledge repository.

Static analysis

No suspicious patterns detected.