Security audit

OpenClaw Skill Builder (Based on Claude)

Security checks for vulnerabilities and agentic risk

Overview

This skill is a straightforward guide for creating and auditing OpenClaw skills, with no hidden code, persistence, credential handling, or external execution behavior.

Before installing, be aware that this meta-skill may activate on broad skill-development wording. It appears safe, but users creating downstream skills should replace generic trigger examples with specific, narrowly scoped phrases and explicit boundaries.

A.I.G

Vulnerability Patterns

Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

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
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Ae1

High

Category: analysis-evasion
Content: - Use the existing skill format (`SKILL.md` in skill folder)
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected

Vague Triggers

Medium

Confidence: 95% confidence
Finding: The trigger phrases are broad and overlap with normal development or review activities such as creating, improving, or auditing skills. In an agentic environment, this can cause the meta-skill to activate unintentionally, steering unrelated workflows and potentially influencing how other skills are created or modified without explicit user intent.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The example guidance encourages phrases like 'do X' and 'help with Y', which are overly generic and can propagate poor triggering practices into newly generated skills. Because this is a meta-skill used to create other skills, the weakness can spread across the workspace and increase accidental invocation or scope confusion in multiple downstream skills.

Static analysis

No suspicious patterns detected.