跳到技能内容
词映 AI导航
发现工具选型指南Skills 技能AI 快讯AI图片AI视频AI文档AI办公AI助手AI搜索AI编程AI音频AI设计AI智能体模型与开发AI学习模型评测AI提示词内容检测AI金融AI 3D扫描与识别其他工具跨境电商 AI
词映 AI导航
工具库Skills 技能
我的收藏
词映 AI导航
发现工具选型指南Skills 技能AI 快讯AI图片AI视频AI文档AI办公AI助手AI搜索AI编程AI音频AI设计AI智能体模型与开发AI学习模型评测AI提示词内容检测AI金融AI 3D扫描与识别其他工具跨境电商 AI
Skills directory/NaNathan13/the-forge/research

Claude Code research for repository-first technical questions

Read the code for uncertain technical questions, then verify with targeted sources as needed. Get agreement before deeper parallel research.

中文

Install in Claude Code

Use Claude Code with the question and codebase. Provide suitable public sources when external research is needed; deeper research uses the available agents.

  1. Review `.claude/skills/research/SKILL.md` and its requirements at the pinned GitHub commit.
  2. Create `.claude/skills/research/` in your project and save the reviewed file as `.claude/skills/research/SKILL.md`.
  3. Start Claude Code from that project's root, then select `/research` in the slash menu.

Example task

The research-28 question asks whether a local filter is persisted. Existing code shows {storage: 'session', key: 'event-order'} and no cross-device sync implementation was found.

What to check

Return {key: 'event-order', persistsWhere: 'session', crossDevice: 'not found'}, the file inspected, and remaining unknowns. Finish the light research with the available evidence.

Original SKILL.md on GitHub · Updated 2026-11-14

Frequently asked question

When should I use parallel research?

Propose subquestions and get agreement when the existing code and a single source are insufficient. Summarize findings with evidence and source locations.

Browse Claude Code skills in the Chinese directory
词映 AI 导航Skills 技能
工具分类全部工具已本地化技能选型指南AI 快讯