dan-research-skillkit

dan-research-skillkit

Academic skills and knowledge packs for researchers, designed for both Claude Code and Codex.
一个同时适用于 Claude Code 与 Codex 的科研技能工具箱，面向硕士生、博士生与研究写作者。

• Dual-host ready: .claude/skills/ and .agents/skills/
• Bilingual skills and knowledge packs
• Checklists, manifests, and build scripts for academic work

Quick links:

• Quick Start
• Chinese README

What This Is / 项目简介

dan-research-skillkit is a bilingual research workflow toolkit. It turns a long-term academic workflow library into reusable dan-* skills, knowledge packs, manifests, and build scripts.

这个仓库把长期积累的科研方法材料整理成可调用的技能、知识包、清单和构建脚本，方便在 Claude Code 与 Codex 中直接使用。

Who This Is For / 适合谁

• Master's students who need a structured research workflow
• PhD students who need stage diagnosis, output planning, and paper production support
• Research assistants, academic writers, and independent scholars
• Anyone who wants skills for literature work, question formation, theory, design, writing, review papers, submission, defense, and long-term planning

What You Get / 你会得到什么

• dan-* topic skills for literature work, research framing, design, writing, review papers, submission, defense, and planning
• topic-based knowledge packs with concepts, frameworks, and checklists
• dual-host layouts for Claude Code and Codex
• manifests, coverage catalogs, and validation scripts for ongoing maintenance
• local extraction and build tools for expanding the toolkit over time

Quick Start / 快速开始

1. Clone the repository.
2. Read docs/quickstart.md.
3. Pick a skill from docs/skills.md.
4. Use repo-local skills directly, or install them globally for Claude Code or Codex.
5. Rebuild generated layouts with python3 scripts/build_skills.py && python3 scripts/build_host_layouts.py.

Install for Claude Code and Codex / Claude Code 与 Codex 安装

For Claude Code, see docs/install-claude.md.
For Codex, see docs/install-codex.md.

Choose A Workflow / 按科研情境选择工作流

Research Workflow Directory / 科研工作流目录

| Skill | Use it when | Typical output | | --- | --- | | dan-research | you need stage diagnosis, routing, or next-output planning | stage note, next-output list, routing decision | | dan-literature | you need search terms, reading layers, or a literature matrix | keyword grid, screening logic, literature workflow | | dan-literature-stack | you need Google Scholar/CNKI acquisition, Zotero project setup, and review-ready corpus prep | stack install report, collection schema, review-ready corpus checklist | | dan-question-gap | you need a tighter research question or contribution claim | question statement, gap note, contribution ladder | | dan-theory | you need a conceptual framework or theory contribution | concept map, mechanism statement, theory paragraph | | dan-design | you need to align question, data, evidence, and analysis | design alignment table, method-fit note | | dan-paper-writing | you need paper structure or section-level writing logic | outline, section interface note, writing checklist | | dan-review-paper | you need a review-paper workflow, not just a literature review | review type decision, screening plan, synthesis frame | | dan-submission | you need submission materials or reviewer-response logic | submission checklist, comment matrix, response skeleton | | dan-defense | you need presentation, proposal, or defense preparation | talk outline, slide logic, defense checklist | | dan-planning | you need pacing, milestones, or execution repair | milestone plan, weekly rhythm, recovery note | | dan-method-atlas | you need cross-method comparison before committing | method comparison table, evidence map | | dan-academic-tools | you need templates, toolchains, or AI-use boundaries | tool stack, template list, workflow support note | | dan-upgrade | you maintain the toolkit and need rebuild/release discipline | update checklist, rebuild plan, release validation |

Use a skill when you need action. Read the linked pack when you need deeper frameworks, pitfalls, and coverage signals.
需要执行时调用 skill；需要更完整的框架、误区和语料覆盖线索时，再读对应的 pack。

How To Work With A Skill / 一个 skill 应该怎么用

1. Start with dan-research when you are unsure which bottleneck matters most.
2. Move into one domain skill and stay there long enough to produce a visible output.
3. Open the matching pack under knowledge/packs when you need more depth.
4. Hand off to the next dan-* skill only after the current output is clear.
5. Use docs/skills.md as the full routing directory.

Literature Stack / 文献采集与综述工作流栈

Use dan-literature-stack when you want one orchestrator for:

• Google Scholar discovery
• CNKI search and export
• Zotero MCP setup and project collections
• batch organization, tagging, and duplicate cleanup
• review-ready source preparation before dan-review-paper

The setup reference lives at skills-src/dan-literature-stack/references/stack-setup.md, and the installer script is scripts/install_literature_stack.py.

Knowledge Packs Overview / Knowledge Packs 总览

Knowledge packs live under knowledge/packs. Each pack includes:

• bilingual summary
• core concepts
• frameworks
• checklists
• template names
• related skills and pack metadata

Repository Structure and Build Flow / 仓库结构与生成方式

• skills-src/: single source of truth for all dan-* skills
• .claude/skills/: generated Claude Code layout
• .agents/skills/: generated Codex-compatible layout
• knowledge/: public packs and manifests
• catalog/: coverage indexes, pack maps, and release backlogs
• scripts/: inventory, extraction, normalization, build, and validation tools
• schemas/: JSON schemas for packs, skills, and source cards

See docs/architecture.md.

Usage Patterns and Troubleshooting / 常见用法与故障排查

Typical usage patterns:

• route a research problem to the right dan-* skill
• read one pack deeply before drafting a chapter or proposal
• vendor repo-local skills into a project workspace
• rebuild both host layouts after adding or revising a skill

Troubleshooting docs:

• docs/install-claude.md
• docs/install-codex.md
• docs/extraction-pipeline.md
• docs/collection-status.md

Roadmap, License, and Author / 路线图、许可与作者

Roadmap:

• expand OCR coverage for image-heavy PDFs in the private extraction layer
• deepen templates and pack examples
• improve similarity checks and source refresh automation

Licenses:

• Code and scripts: LICENSE
• Documentation and packs: LICENSE-docs

Author / 作者：南洋理工DAN读博笔记（小红书）