搜索

PythonReactSEOPDF
GitHub Skill

judokon-planning-and-investigation

Supports planning, investigation, and analysis tasks by encouraging structured reasoning and use of the JU-DO-KON! RAG vector database.

CyanAutomation☆ 1 仓库 Stars0 仓库 Forks

安装与使用

请根据来源仓库和你的 AI 工具说明安装此技能。先阅读技能文件和依赖要求,再添加到工作流。

查看来源 ↗使用指南

技能说明

Skill Instructions

This skill is for thinking before coding.

Inputs / Outputs / Non-goals

  • Inputs: user questions, relevant PRDs/docs, RAG queries and results.
  • Outputs: investigation summary, sources, unknowns, recommended next steps.
  • Non-goals: implementation changes without a follow-up task.

What this skill helps accomplish

  • Avoid reinventing solutions
  • Surface existing knowledge
  • Ground decisions in prior work

When to use this skill

  • Investigating bugs
  • Planning new features
  • Exploring design options
  • Answering “how does this work?” questions

Mandatory behaviour

  1. Search the RAG vector database first
  2. Summarise relevant findings
  3. Cite existing code, PRDs, or docs
  4. Identify unknowns explicitly

RAG format

  • Query template: "topic + file type + context" (example: "battle timer phases scoreboard implementation").
  • Prefer withProvenance: true and quote sources in the summary.

Investigation steps

  1. Query RAG embeddings for related concepts
  2. Review similar past implementations
  3. Identify constraints and invariants
  4. Propose next steps (not solutions yet)

Stop conditions

  • If RAG results are weak twice, fallback to targeted file search.

Expected output

  • Investigation summary
  • Referenced sources
  • Clear recommendations or questions

使用示例

查看来源仓库中的使用示例与配置说明。

查看来源 ↗

来源历史

收录时间
2026-01-20
索引更新时间
2026-01-26
Branch
main

收录与同步日期属于本站记录。源码修改历史请查看来源仓库。

GitHub ↗

chat Comments (0)

chat_bubble_outline

No comments yet. Be the first to share your thoughts!