技能 repro-first-debugging
📦

repro-first-debugging

v1.0 内容修订 r1 安全 ⚙️ 外部命令

Debug Failures With Reproducible Evidence

Bug fixes based on guesses can miss the cause and introduce regressions. This skill guides reproduction, hypothesis testing, focused fixes, and verification with recorded evidence.

支持: Claude Codex Code(CC)
你是维护者？认领此 Skill 🥉 78 青铜

使用我的智能体安装

将此请求复制给你的智能体，其中包含权威 Skill 页面和清单。

智能体请求
Review the Skillstore skill "repro-first-debugging" from https://skillstore.io/skills/agentkitco-repro-first-debugging.md and its manifest at https://skillstore.io/api/skills/agentkitco-repro-first-debugging/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.

你的智能体仍应展示其计划，并按安全策略请求必要确认。

Agent 可读资源

当 AI Agent、爬虫或脚本需要干净上下文，而不是读取完整页面时，请使用这些链接。

Markdown 详情 GET /skills/agentkitco-repro-first-debugging.md 已签名 manifest GET /api/skills/agentkitco-repro-first-debugging/manifest 已签名 lockfile GET /api/skills/agentkitco-repro-first-debugging/lockfile

测试它

正在使用“repro-first-debugging”。 Illustrative request: investigate a parser crash when the input is empty.

预期结果:

  • Failure: Empty input should return a validation error, but parsing crashes.
  • Reproduction: The smallest empty-input test reproduces the crash.
  • Root cause: The parser reads the first item without checking whether the input contains any items.
  • Evidence: The empty-input case fails before the change; a nonempty case succeeds.
  • Fix: Validate empty input before accessing the first item.
  • Verification: The original reproduction and nearby parser tests pass after the change.
  • Related risks: Similar unchecked access was reported for separate review.

正在使用“repro-first-debugging”。 Illustrative request: fix an intermittent crash that cannot be reproduced locally.

预期结果:

  • Failure: A reported intermittent crash remains unconfirmed in the available environment.
  • Reproduction: The supplied steps did not trigger the failure.
  • Fix: No implementation changes made.
  • Needed evidence: Exact error text, triggering input, environment details, and a repeatable failing command.

安全审计

安全
v1 • 2026/10/17 打开版本化报告

All four static findings are false positives involving Markdown formatting, legitimate debugging guidance, or a goal statement. No evidence found of malicious command execution, system reconnaissance, exfiltration, or prompt injection in SKILL.md.

1
已扫描文件
43
分析行数
0
复核项
0
已忽略误报

风险因素

⚙️ 外部命令 (3)
SKILL.md:20 SKILL.md:23 SKILL.md:27-35
最新一次完成的静态与语义审计未检测到已确认的安全发现。这并不能证明该 Skill 没有副作用。
审计者: codex
分享与引用此报告

分享版本化评估报告、中性徽章、嵌入卡片和引用。Skillstore 只报告证据，不判断该 Skill 是否安全。

打开版本化报告
安全评估

复制报告链接

https://skillstore.io/skills/agentkitco-repro-first-debugging/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown 徽章

[![Skillstore security assessment](https://skillstore.io/badges/skills/agentkitco-repro-first-debugging/security.svg)](https://skillstore.io/skills/agentkitco-repro-first-debugging?utm_source=security_passport_badge)

HTML 徽章

<a href="https://skillstore.io/skills/agentkitco-repro-first-debugging?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/agentkitco-repro-first-debugging/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

嵌入卡片

<iframe src="https://skillstore.io/embed/skills/agentkitco-repro-first-debugging.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>
学术引用 (APA · BibTeX · CFF)

APA 引用

agentkitco. (2026). repro-first-debugging security audit report (audit version 1) [Author version 1.0]. Skillstore. https://skillstore.io/skills/agentkitco-repro-first-debugging/audits/1

BibTeX 引用

@techreport{agentkitco-agentkitco-repro-first-debugging-2026, author = {agentkitco}, title = {repro-first-debugging security audit report (audit version 1)}, institution = {Skillstore}, year = {2026}, number = {1}, url = {https://skillstore.io/skills/agentkitco-repro-first-debugging/audits/1}, note = {Author version 1.0} }

CITATION.cff

cff-version: 1.2.0 message: "If you use this Skill, cite its author and this versioned security audit report." title: "repro-first-debugging security audit report (audit version 1)" version: "1.0" type: report authors: - name: "agentkitco" date-released: "2026-10-17" url: "https://skillstore.io/skills/agentkitco-repro-first-debugging/audits/1" identifiers: - type: other value: "skillstore:agentkitco-repro-first-debugging:audit:1" description: "Skillstore immutable audit report identifier"

Skillstore 评分

评分依据 证据置信度: 中
55
架构
85
可维护性
87
内容
65
社区
91
规范符合性

你能构建什么

Investigate an application crash

Reduce a reported crash to a small reproduction, test plausible causes, and verify a focused correction.

Turn a defect report into a regression test

Clarify expected behavior and preserve a failing scenario as an automated test before validating the repair.

Resolve a repository regression

Inspect recent changes, collect evidence for competing hypotheses, and report similar patterns without expanding the patch.

试试这些提示

Begin with a clear failure 
Use repro-first-debugging to investigate [failure]. Expected behavior: [expected]. Actual behavior: [actual]. Ask for missing details before attempting reproduction.
Minimize a failing test 
Reproduce [test failure] with [test command]. Capture its output and reduce the failing input. Prefer an automated regression test before changing implementation.
Compare root cause hypotheses 
Investigate [reproducible failure]. Rank three hypotheses by likelihood and checking cost. Test them and record evidence. Stop if all three fail.
Verify a regression repair 
Reproduce [regression] using [command]. Inspect recent changes and test hypotheses. Apply the smallest supported fix. Show before-and-after results and report related patterns without edits.

最佳实践

  • Provide exact error text, expected behavior, and the smallest known reproduction command.
  • Preserve the failing scenario as a regression test and record evidence for each hypothesis.
  • Keep fixes focused, remove temporary logging, and include before-and-after verification results.

避免

  • Editing implementation before reproducing the failure without an explicit request for a blind fix.
  • Deleting or weakening tests to make verification pass.
  • Claiming success without before-and-after command evidence or changing unrelated code without asking.

常见问题

What does this skill provide?
It provides a debugging workflow and report structure centered on reproduction, hypothesis testing, focused fixes, and verification.
Which assistants can use it?
The marketplace lists Claude, Codex, and Claude Code as supported tools.
What happens when the failure cannot be reproduced?
The workflow stops and reports attempted reproductions. Implementation changes require successful reproduction unless you explicitly request a blind fix.
Does it include scripts or testing tools?
No. The reviewed package contains only SKILL.md and relies on the assistant and existing project tools.
How does it verify a fix?
It reruns the original failing test and nearby tests, then the full suite when fast. The report includes before-and-after evidence.
When does hypothesis testing stop?
After three failed hypotheses, the workflow stops and reports current evidence and missing information.

开发者详情

作者

agentkitco

许可证

MIT

作者版本

v1.0

Skillstore 修订

r1

版本提示

作者声明的版本不符合 SemVer 规范。

仓库

https://github.com/agentkitco/agentkit-dev-skills/tree/3de707c7875147b65ddff504d86b0d0dbf713fa7/skills/repro-first-debugging

引用

743429f75b17204d47661a86b577a575ad67517f

维护新鲜度

2026/10/17

使用量

0 次下载 · 0 次浏览

文件结构

📄 SKILL.md

agentkitco 的更多技能

查看全部

相关技能

查看全部