cms-pua-skill
PassAudited by ClawScan on May 10, 2026.
Overview
This instruction-only skill is openly designed to make the agent more persistent, but it does not install code or request credentials; users should still set boundaries around commands and scope.
This skill is reasonable if you want the agent to be more persistent during debugging or problem solving. Before using it, set clear limits: what files it may inspect, what commands it may run, when it must ask before acting, and when it should stop and report uncertainty.
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 keep trying longer before admitting limits or asking for help.
This changes the agent’s normal stopping/reporting behavior by discouraging early admission of inability until it has tried major paths.
穷尽主要方案前，不得说“我无法解决”。
Use this skill when you want persistence, but give explicit stop conditions and remind the agent not to exceed user approvals or safety boundaries.
The agent may inspect project files or run diagnostic commands more readily during debugging tasks.
The skill encourages proactive searches, file reads, and command execution before asking the user, which is purpose-aligned for troubleshooting but can broaden tool use.
先做后问。优先搜索、读文件、跑命令、查上下文；只有确实缺少用户独有信息时才提问
Limit the agent to the intended project/workspace and require confirmation for destructive, privileged, network-heavy, or broad file-system actions.
The agent may sound more forceful and persistent than usual, which could make its progress claims feel more confident.
The skill uses high-pressure framing to push agent behavior. This is disclosed and aligned with the skill’s purpose, but it can reduce caution if not bounded.
用高压话术加系统化方法，逼自己穷尽方案
Ask for objective evidence, intermediate status, and candid uncertainty; do not treat persistence as proof that the final answer is correct.