Jue AI Awakening Engine
PassAudited by VirusTotal on May 5, 2026.
Overview
Type: OpenClaw Skill Name: jue-zh Version: 1.0.0 The skill bundle is a meta-cognitive framework designed to improve AI agent reliability and self-correction during task failures. It uses philosophical metaphors (e.g., 'Three Poisons', 'Seven Jue') to instruct the agent to stop repeating failed patterns, verify assumptions using standard debugging tools (like ls, which, and curl), and provide structured status reports. There is no evidence of malicious intent, data exfiltration, or harmful prompt injection; the instructions are focused entirely on quality-of-service improvements and logical troubleshooting.
Findings (0)
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.
The agent may become more persistent and proactive during troubleshooting, which can be helpful but may reduce opportunities for user confirmation.
The skill defines mandatory invocation and stopping behavior and encourages best-guess action instead of waiting for user direction.
都必须触发此技能... 清单未完成，不能说"我无法解决" ... 等用户指示下一步 | 你比用户更清楚下一步 | 先做最佳猜测
Use this skill with normal approval rules intact, especially for changes that modify files, call external services, spend money, or affect accounts.
In environments with powerful tools, the agent may run diagnostic commands or searches before asking additional questions.
The skill instructs the agent to use tools, inspect source/context, run commands, and perform diagnostic network-style checks as part of validation.
使用工具验证假设... 搜索、读源码、跑命令 ... build / test / curl，**贴输出**
Keep tool permissions scoped and require confirmation for destructive, sensitive, costly, or externally visible actions.
Diagnostic notes could carry stale assumptions or task details into later work if used in a multi-agent environment.
The multi-agent guidance describes retaining and reusing troubleshooting records across teammates and task handoffs.
维护全局照见记录（按 teammate + 任务维度）... 任务从 A 重分配给 B 时附带照见记录
Keep these records task-scoped, review them before reuse, and avoid storing secrets or unnecessary private data in troubleshooting summaries.
If used with multiple agents, task context may be shared among teammates or broadcast more widely than a single-agent conversation.
The reference file defines inter-agent state transfer and broadcast channels for sharing troubleshooting context.
Leader → Teammate | 任务描述 + `Teammate write` ... Leader → All | `broadcast` ... Mirror → Teammate | `Teammate write`
Use only trusted agents/channels and avoid broadcasting sensitive task details unless necessary.