Back to Details

spike

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Spike

Spike

Structured technical investigation to reduce uncertainty. Answer specific questions, not "explore X."
结构化技术调研，用于降低不确定性。要回答具体问题，而非「探索X」这类泛泛的需求。

Key Principles

核心原则

  • Every spike answers a specific, measurable question
  • Strict time-box (default 4h) — produce a decision at the end, even if incomplete
  • Output is a decision (GO/NO-GO/PIVOT/MORE-INFO), not a general exploration
  • Create follow-up tasks from findings, not just a report
  • 每个spike都要回答一个具体、可衡量的问题
  • 严格的时间限定（默认4h）—— 即使调研不完整，到期也要输出决策
  • 输出是决策（GO/NO-GO/PIVOT/MORE-INFO），而非泛泛的探索结果
  • 基于调研发现创建后续任务，而非仅输出报告

When NOT to Use

不适用场景

  • Multi-day evaluations — Use 
    deep-research
    for comprehensive technology evaluations with stakeholder reports
  • Already know what to build — Skip straight to implementation; spikes are for reducing uncertainty, not planning known work
  • Bug investigation — Use 
    error-handling
    or 
    /fix
    ; bugs have reproduction steps, spikes have open questions
  • 多日评估 —— 需开展全面技术评估并输出面向利益相关者的报告时，请使用
    deep-research
  • 已明确开发内容 —— 直接进入实现环节即可；spike用于降低不确定性，而非规划已知工作
  • Bug排查 —— 请使用
    error-handling
    或
    /fix
    ；Bug有明确的复现步骤，而spike针对的是开放性问题

Quick Start Checklist

快速入门检查清单

  1. Define the question clearly (what are we trying to learn?)
  2. Set a time-box (hours, not days — use deep-research for multi-day)
  3. Identify evaluation criteria upfront
  4. Investigate with focused experiments or prototypes
  5. Checkpoint at 50% — assess progress, decide if pivoting
  6. Produce mandatory conclusion: GO / NO-GO / PIVOT / MORE-INFO
  1. 明确定义调研问题（我们想要了解什么？）
  2. 设置时间限定（以小时为单位，不要按天设置——多日调研请使用deep-research）
  3. 预先明确评估标准
  4. 通过针对性实验或原型开展调研
  5. 时间过半时进行进度检查——评估进展，决定是否需要调整方向
  6. 必须输出明确结论：GO / NO-GO / PIVOT / MORE-INFO

Mandatory Outputs

强制输出项

  • Decision: GO, NO-GO, PIVOT, or MORE-INFO
  • Evidence: What was tested, results observed
  • Follow-up tasks: Created in ohno if GO
  • Report: Saved to 
    .claude/spikes/[name]-[date].md
  • 决策：GO、NO-GO、PIVOT 或 MORE-INFO
  • 证据：测试内容、观测到的结果
  • 后续任务：若决策为GO，需在ohno中创建
  • 报告：保存至 
    .claude/spikes/[name]-[date].md

References

参考文档

ReferenceDescription
spike-types.mdFeasibility, architecture, integration, performance spikes
question-patterns.mdHow to frame good spike questions
output-templates.mdSpike report templates and examples
参考链接描述
spike-types.md可行性、架构、集成、性能类spike说明
question-patterns.md如何设计合理的spike调研问题
output-templates.mdSpike报告模板及示例