ClawHub
Back to skill

Security audit

Knowledge Graph Extractor

Security checks for vulnerabilities and agentic risk

Overview

This appears to be a document-processing skill whose JSON editing behavior is purpose-aligned, with no evidence of malware, exfiltration, or hidden persistence.

Installers should understand that this skill may modify generated JSON files during its quality-check step. Review the resulting diff or keep a backup when accuracy matters, especially for educational or structured knowledge outputs.

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (3)

Missing User Warnings

Medium
Confidence
85% confidence
Finding
The skill instructs the agent to automatically modify generated JSON files in a loop without explicit user confirmation or a clear audit boundary. In a document-processing workflow, this can silently alter extracted educational content, fabricate missing hierarchy/objectives, or overwrite user-expected output, creating integrity and trust risks even if the intent is quality improvement.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The prompt explicitly instructs the agent to use an Edit tool to modify the JSON file in place as soon as issues are found, but it provides no requirement to warn the user, confirm consent, or create a backup first. In a file-processing skill, this can lead to silent alteration of user data, including overcorrection or destructive changes to extracted knowledge structures.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The execution flow normalizes repeated read/edit/re-read cycles on the target JSON file without disclosure that the agent will continuously modify the original file. This increases the chance of unintended cumulative changes, data loss, or hard-to-audit edits, particularly because the loop encourages autonomous rewriting until the prompt's criteria are satisfied.

Static analysis

No suspicious patterns detected.