ClawHub
Back to skill

Security audit

skill-factory

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed skill-building helper that can create new skill packages, with some guardrails around secrets and overwrites.

Install this only if you intend to use an agent to draft or scaffold new OpenClaw skills. Review every generated skill before enabling it, especially its trigger conditions, tool permissions, credential handling, file-write behavior, and whether it could persist unsafe instructions. Be careful when generating skills for high-impact domains or anything that can modify accounts, money, public content, or private data.

SkillSpector

Vulnerability Patterns
  • Rogue AgentSelf-Modification, Session Persistence
  • 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
Findings (4)

Vague Triggers

Medium
Confidence
89% confidence
Finding
The activation examples are very broad ('make a skill for X') and can overlap with ordinary user requests, causing the meta-skill to trigger in situations where the user only wanted direct assistance. In a skill that generates other skills, this broad triggering increases the chance of unintended capability escalation and creation of unsafe or over-permissive downstream skills without explicit user consent.

Vague Triggers

Medium
Confidence
83% confidence
Finding
The template is designed to generate arbitrary new skills 'from scratch' for any domain, but it does not constrain invocation scope, allowed capabilities, or prohibited behaviors. In a meta-skill context, overly broad creation guidance can propagate unsafe patterns into downstream skills, including skills with excessive authority, weak safety boundaries, or ambiguous triggering criteria.

Self-Modification

High
Category
Rogue Agent
Content
→ Output system design

### Fase 3: Core Building
→ Write SKILL.md (the brain)
→ Define purpose, triggers, directives
→ Install reasoning protocols
→ Set quality standards
Confidence
92% confidence
Finding
The skill is explicitly designed to generate new SKILL.md files, including purpose, triggers, directives, and reasoning protocols for other skills. In a meta-skill context, this is effectively self-modification or capability propagation: it can manufacture new agent behaviors, potentially embedding unsafe instructions, bypass patterns, or over-broad triggers into generated skills that are then activated elsewhere.

Self-Modification

High
Category
Rogue Agent
Content
→ Design the output system

PHASE 3: CORE BUILDING
  → Write SKILL.md (the brain)
  → Define purpose, triggers, directives
  → Install reasoning protocols
  → Set quality standards
Confidence
95% confidence
Finding
This skill explicitly instructs the agent to write SKILL.md and package a new skill, which is a self-modification/code-generation capability affecting future agent behavior. In a skill ecosystem, generating or overwriting skill files can persist unsafe instructions, broaden authority, or introduce prompt-injection/backdoor logic that survives beyond the current session.

Static analysis

No suspicious patterns detected.