Claw Ethics Checker

PassAudited by ClawScan on May 10, 2026.

Overview

This appears to be a local advisory ethics/compliance checker, but users should not treat its simple rule checks as legal advice and should handle decision logs carefully.

Before installing, verify the package source, keep human review enabled for meaningful decisions, and avoid entering secrets or sensitive personal data into task details that may be logged or exported.

Findings (4)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

NoteHigh Confidence

ASI06: Memory and Context Poisoning

What this means

Sensitive task details could be retained in logs or exported if the user or an integrating agent enables that workflow.

Why it was flagged

The skill records full task descriptions and task details in a decision log, which may include sensitive user or business information.

Skill content

log_entry = {
            'task_description': task_description,
            'task_details': task_details,
            'result': result.to_dict(),
            'timestamp': self._get_timestamp()
        }
        self.decision_log.append(log_entry)

Recommendation

Keep logs local, avoid logging secrets or private data, and review any exported decision-log files before sharing them.

NoteHigh Confidence

ASI01: Agent Goal Hijack

What this means

If integrated as shown, the skill can influence whether an agent proceeds with, rejects, or escalates a user task.

Why it was flagged

The examples show the checker being used to reject or escalate tasks based on its risk classification.

Skill content

if ethics_result.risk_level == RiskLevel.HIGH:
            return {
                'status': 'rejected',
                'reason': '高风险任务，基于伦理检查拒绝'

Recommendation

Use it as an advisory guardrail with clear thresholds and human review, especially for important legal, business, or safety decisions.

NoteMedium Confidence

ASI09: Human-Agent Trust Exploitation

What this means

Users might assume the checker provides authoritative legal compliance decisions when it is only a simple rule-based aid.

Why it was flagged

The implementation states that it currently uses built-in rules, indicating that its compliance judgments are limited and should not be over-trusted.

Skill content

# 这里可以连接外部数据库或加载本地规则文件
        # 目前使用内置规则

Recommendation

Do not rely on this skill as legal advice; keep human/legal review for consequential or ambiguous tasks.

NoteMedium Confidence

ASI04: Agentic Supply Chain Vulnerabilities

What this means

Installing from a different remote source could expose users to code that differs from the reviewed artifact.

Why it was flagged

The documentation includes user-directed installation from external repositories/download URLs, while the registry source is unknown.

Skill content

git clone https://github.com/openclaw/claw-ethics-checker.git
# 或直接下载
wget https://github.com/openclaw/claw-ethics-checker/archive/main.zip

Recommendation

Prefer the registry package or a verified repository, and review downloaded code before installing or running it.