Security audit

JiaHeng Asset Architect

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed companion persona for building and pre-checking OpenClaw platform assets, with broad but purpose-aligned file and command permissions.

Install this if you want a Chinese-language, strict pre-review persona for OpenClaw asset creation alongside wb-open-builder. Be aware it may activate on broad asset-building requests and uses file-writing plus Bash authority for generation, validation, and packaging, so invoke it in a workspace where those actions are expected.

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

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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)

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The trigger list includes broad phrases such as “开放平台伴生人格” and references to paired-loading with another skill, which can cause this skill to activate in conversations that merely discuss the related tool or general platform-review topics. Unintended activation can override user expectations, alter response behavior, and introduce prompt-scope interference with other concurrently loaded skills.

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The statement that the persona ‘automatically takes effect’ whenever a user raises an open-platform asset request is overly broad for shared sessions and creates implicit activation without explicit consent. In multi-skill or general-purpose conversations, this can unexpectedly steer outputs, enforce hidden workflow constraints, and create prompt-priority conflicts that are hard for users to detect.

Natural-Language Policy Violations

Medium

Confidence: 88% confidence
Finding: Forcing Simplified Chinese as the response style without user opt-in can conflict with the user’s language preference and reduce transparency or usability, especially in multilingual contexts. While not a direct code-execution risk, it is a behavioral control issue that can degrade accessibility, cause misunderstandings, and make other instructions harder to follow accurately.

Static analysis

No suspicious patterns detected.