ClawHub
Back to skill

Security audit

mb-text-workflow

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed markdown documentation workflow for maintaining project memory-bank files, with minor scoping notes but no evidence of hidden or malicious behavior.

Install this only in projects where you want the agent to maintain a text-based memory bank. Expect it to read project history/status and update markdown files under memory-bank; if you require approval before every new file, clarify that before using the workflow.

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Intent-Code Divergence

Medium
Confidence
96% confidence
Finding
The session management workflow instructs the agent to create missing session files automatically, which conflicts with earlier explicit rules forbidding file modifications without user approval. In an agent skill, contradictory instructions like this can bypass the intended human approval gate and lead to unauthorized file creation or state changes.

Intent-Code Divergence

Medium
Confidence
97% confidence
Finding
This workflow again tells the agent to create a missing session file as part of the memory bank update process, despite repeated earlier instructions that no files may be modified without explicit approval. Such inconsistency is dangerous because agents often follow the most specific procedural step, resulting in unapproved writes to the workspace.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The trigger list includes generic phrases such as "update memory bank," "update tasks.md," and "update session cache," which can activate the skill in a wide range of ordinary editing contexts. Overly broad activation increases the chance the agent applies this workflow when the user did not intend a memory-bank operation, potentially causing unnecessary file reads/edits and unintended documentation changes.

Static analysis

No suspicious patterns detected.