PUA Debugging
ReviewAudited by ClawScan on May 10, 2026.
Overview
This skill has no code or credentials, but it pressures the agent to keep going and use tools broadly before asking you, which could push it outside your intended scope.
Install only if you intentionally want a highly proactive, pressure-driven debugging prompt. Use it with approval gates for shell, file, and network tools, and tell the agent not to inspect unrelated files, run side-effecting commands, or continue indefinitely without your confirmation.
Findings (3)
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 enter this aggressive debugging mode in many situations and continue acting when you expected it to pause, ask, or admit limits.
The skill makes itself broadly authoritative based on failures or frustration, not only explicit user requests, and changes the agent's normal trigger and stopping behavior.
“MUST trigger when ... user expresses frustration in ANY form ... Applies to ALL task types”
Limit activation to explicit debugging tasks and preserve normal user-approval, safety, and stop conditions.
If the agent has local or remote tools, it may inspect more data or run more commands than you intended before asking for clarification.
The skill mandates use of search, file-reading, and command-execution tools before asking the user, without clear limits on commands, files, environments, approvals, or reversibility.
“你有搜索、文件读取、命令执行等工具。在向用户提问之前，必须先用工具自行排查。”
Keep per-action approval enabled for shell, file, and network tools; require the agent to ask before inspecting unrelated files or running commands with side effects.
The agent may sound more forceful or overconfident and may delay telling you that a task needs human input.
The skill openly uses coercive motivational language to discourage giving up; this is disclosed, but it may reduce candid uncertainty or appropriate escalation.
“用中西大厂 PUA 话术让你不敢放弃”
Use this only when you want an aggressive debugging style, and override it when honesty, caution, or explicit confirmation is more important.