ClawHub
Back to skill

Security audit

Human Approval

Security checks for vulnerabilities and agentic risk

Overview

This is a transparent, instruction-only approval reminder that asks before risky actions and does not contain hidden code or data access.

Install this only as a convenience reminder, not as a security boundary. It relies on the model to follow the instructions, can be disabled with `/human-approval off`, and the skill itself warns that prompt injection, tight loops, or context loss can bypass it.

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 (1)

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
### `/human-approval off`

Temporarily disable approval prompts for the current session.

## Limitations
Confidence
95% confidence
Finding
The skill explicitly provides a `/human-approval off` command that disables approval prompts for the current session, removing the only control this skill is meant to provide. Because the skill itself acknowledges it is only a soft, context-window-based gate, this disable switch can be abused or socially engineered to bypass human review before destructive actions.

Static analysis

No suspicious patterns detected.