← 返回 Skills 市场
doraemon-claw

pua-debugging-pro

作者 Doraemon-Claw · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ✓ 安全检测通过
709
总下载
1
收藏
3
当前安装
1
版本数
在 OpenClaw 中安装
/install pua-debugging-pro
功能描述
Professional anti-giveup debugging protocol for coding tasks where the agent starts looping, deflecting to users, or trying to end early without evidence. Us...
使用说明 (SKILL.md)

PUA Debugging Pro (Dignified Edition)

Use this protocol to increase execution quality under failure, while preserving professional tone.

Non-negotiables

  1. No premature surrender

    • Do not conclude "cannot solve" before completing escalation checklist.
  2. Evidence before questions

    • Use available tools first.
    • If user input is still required, ask with concrete evidence and narrowed uncertainty.
  3. Verification before completion

    • No "done" claims without explicit validation output.
  4. Dignified communication

    • Never use humiliation or threatening rhetoric.
    • Use calm, direct, engineering language.

Trigger signals

Activate when one or more are present:

  • 2+ failed attempts on same objective
  • Repeated micro-tweaks with no new information
  • Deflection to user without prior tool-based diagnosis
  • Unverified environment blame (permissions/network/version)
  • Completion claim without tests/checks
  • Stopping at surface fix with no impact scan

Escalation ladder (bounded)

L1 (after 2 failed attempts)

  • Switch to a substantially different hypothesis.
  • Record: what failed, what changed, what signal to watch.

L2 (after 3 failed attempts)

  • Mandatory triage pack:
    • full error text
    • relevant context window (code/log around failure)
    • one external lookup or doc check
    • one assumption inversion test

L3 (after 4+ failed attempts)

  • Run full 7-point checklist (below).
  • Produce structured decision: continue / pivot / stop.

7-point checklist

  • Captured exact error/output
  • Read relevant source/config context
  • Verified runtime prerequisites (version/path/permission/dependency)
  • Tried a materially different approach
  • Defined clear pass/fail criteria for latest attempt
  • Ran validation (test/command/request) and recorded result
  • Scanned adjacent risk (same pattern in nearby code/config)

Output contract

When progressing:

  • Current hypothesis
  • Actions executed
  • Observed evidence
  • Next step

When blocked after L3:

  • Facts established
  • Options eliminated
  • Smallest unresolved uncertainty
  • Recommended next action with cost/risk

For reusable output format, read:

  • references/checklist-template.md

For one-page handoff/retrospective artifact, use:

  • assets/postmortem-onepager.md

Stop conditions (required)

If all conditions are true, stop trying and escalate to user:

  1. 7-point checklist completed
  2. Last attempt produced no new diagnostic signal
  3. Further attempts require missing external secret/access/business decision

Use this closing format:

  • "I completed bounded escalation and cannot safely proceed without X."
  • "Evidence gathered: ..."
  • "Recommended next action: ..."

Style rules

  • Prefer concise, factual updates over motivational talk.
  • Be direct; avoid apology loops.
  • Keep tone firm, respectful, and professional.
安全使用建议
This skill appears coherent and low-risk: it is a protocol and templates to make debugging more evidence-driven. Before installing, confirm that the agent will only be granted the minimal project-level file/log access needed (avoid broad filesystem or secret access). Review the included THIRD_PARTY_NOTICES if you must comply with upstream licensing. If you are concerned about autonomous agent actions, restrict the agent's runtime permissions (filesystem, network, secrets) or require explicit user confirmation before the agent reads sensitive files or runs commands.
功能分析
Type: OpenClaw Skill Name: pua-debugging-pro Version: 1.0.0 The skill bundle provides a structured debugging protocol and reporting templates (SKILL.md, checklist-template.md) designed to improve AI agent persistence and evidence-based troubleshooting. It contains no executable code, network requests, or instructions to access sensitive data or bypass security controls. While it references an upstream project associated with 'PUA' (a slang term for behavioral manipulation of LLMs), this version is explicitly professionalized to enforce engineering standards and dignified communication without any malicious intent.
能力评估
Purpose & Capability
Name/description (anti-giveup debugging) match the SKILL.md content and the included checklist/postmortem templates. There are no unexpected environment variables, binaries, or third-party services required.
Instruction Scope
Instructions appropriately direct the agent to capture errors, read relevant source/config context, run validations, and perform external doc lookups. This naturally requires the agent to access project files, logs, and run commands; that access is coherent with debugging but could expose sensitive local data if the agent has broad filesystem or network permissions. The skill does not instruct exfiltration or use of unknown endpoints.
Install Mechanism
Instruction-only skill with no install spec and no code files to execute. Low risk from install mechanism (nothing is written to disk by the skill itself).
Credentials
No environment variables, credentials, or config paths are requested. The SKILL.md's recommended actions (reading code/logs, checking runtime prerequisites) explain why local access may be needed; those needs are proportionate to the debugging purpose.
Persistence & Privilege
Skill is not always:true and does not request persistent system-wide changes. It is user-invocable and can be invoked autonomously by the agent (platform default) but that is appropriate for a debugging protocol.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install pua-debugging-pro
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /pua-debugging-pro 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
Initial Release: Introduces a structured, dignified debugging protocol to improve problem-solving quality during repeated failures or low-agency situations. - Enforces evidence-driven debugging, verification before task completion, and escalation only after concrete checkpoints. - Provides a multi-level escalation ladder (L1–L3) with clear checklists and required actions after each failure stage. - Mandates dignified, professional tone—no humiliation or threatening language. - Specifies output contracts for progress and blocked states, ensuring transparent communication. - Defines strict stop conditions and closing procedure when further action is not safely possible.
元数据
Slug pua-debugging-pro
版本 1.0.0
许可证 MIT-0
累计安装 3
当前安装数 3
历史版本数 1
常见问题

pua-debugging-pro 是什么?

Professional anti-giveup debugging protocol for coding tasks where the agent starts looping, deflecting to users, or trying to end early without evidence. Us... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 709 次。

如何安装 pua-debugging-pro?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install pua-debugging-pro」即可一键安装,无需额外配置。

pua-debugging-pro 是免费的吗?

是的,pua-debugging-pro 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

pua-debugging-pro 支持哪些平台?

pua-debugging-pro 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 pua-debugging-pro?

由 Doraemon-Claw(@doraemon-claw)开发并维护,当前版本 v1.0.0。

💬 留言讨论