ClawHub
Back to skill

Security audit

知识图谱工具(免费版)

Security checks for vulnerabilities and agentic risk

Overview

This skill appears to be a local knowledge-graph tool, but it needs review because it can persist memory and instructions into agent configuration files with broad activation wording.

Before installing, review exactly what would be written to AGENTS.md, CLAUDE.md, or GEMINI.md, and use it only in a workspace where persistent local memory is intended. Do not store secrets or sensitive private notes in the graph unless you have verified how summaries are generated and what enters future agent context.

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
73% confidence
Finding
The trigger conditions are overly broad and extend beyond the stated scope of a local JSON knowledge-graph tool, encouraging use for generic data analysis, reporting, and visualization tasks. In an agent setting, overbroad activation can cause the skill to run in inappropriate contexts, increasing the chance of unnecessary file access, command execution, or unintended modification of local knowledge data.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The installation instructions state that the script automatically detects the platform and modifies AGENTS.md, CLAUDE.md, or GEMINI.md to inject instructions and summaries, but this side effect is not surfaced as a prominent warning. Modifying agent instruction files changes future model behavior and persistence boundaries, which can create prompt-injection persistence, unauthorized context manipulation, and hard-to-notice policy drift across sessions.

Static analysis

No suspicious patterns detected.