Huashu Research
- 636 installs
- 1.3k repo stars
- Updated August 2, 2026
- alchaincyf/huashu-skills
huashu-research is an agent skill that runs a structured web research workflow with incremental saves to markdown knowledge files so developers do not lose findings when context windows truncate or sessions reset.
About
huashu-research is a structured web research agent skill that persists findings to disk before searching begins. Step 1 creates `_knowledge_base/research-<主题>-<YYYYMMDD>.md` with goals, key questions, and empty findings sections. Step 2 appends each WebSearch result with source URLs immediately. Step 3 writes a stage summary after every 3 searches. Step 4 produces a final briefing. The skill triggers on research, search, or information-gathering requests and follows source-priority rules from SHARED-RULES.md. Developers reach for huashu-research when multi-round web investigation must survive long agent sessions, such as competitive analysis, technology evaluations, or pre-writing research for technical articles.
- Creates a dedicated research markdown file before any search begins
- Appends every discovery with source URL and date in real time
- Generates a stage summary after every 3 searches
- Produces a final structured brief with key facts, credibility ratings, source table and writing suggestions
- Enforces strict separation between research and drafting
Huashu Research by the numbers
- 636 all-time installs (skills.sh)
- Ranked #647 of 3,282 Productivity & Planning skills by installs in the Skillselion catalog
- Data as of Aug 5, 2026 (Skillselion catalog sync)
npx skills add https://github.com/alchaincyf/huashu-skills --skill huashu-researchAdd your badge
Show developers this skill is listed on Skillselion. Paste this into your README.
| Installs | 636 |
|---|---|
| repo stars | ★ 1.3k |
| Last updated | August 2, 2026 |
| Repository | alchaincyf/huashu-skills ↗ |
How do you persist web research across long agent sessions?
Ensure every web research session creates a persistent, incrementally updated knowledge file that survives context windows and session resets.
Who is it for?
Developers running multi-round web research who need findings saved to files before context windows truncate.
Skip if: Single quick factual lookups that do not require a persistent research document or incremental source tracking.
When should I use this skill?
The user requests research, competitive analysis, technology discovery, or multi-search information gathering that must survive session resets.
What you get
Incrementally updated markdown research files with goals, findings, source lists, stage summaries, and a final briefing.
- Research markdown file
- Source URL list
- Final research briefing
By the numbers
- 4-step research workflow from file creation to final briefing
- Stage summaries written after every 3 searches
Files
调研 Skill
结构化的网络调研流程，核心目标：调研成果实时持久化，防止会话截断丢失工作。
何时使用
- 为写文章做前期调研
- 了解新产品、新技术、新发布
- 搜集竞品信息或行业动态
- 任何需要多次 WebSearch 的信息搜索任务
执行流程
Step 1: 立即创建调研文件
- 在开始搜索之前，先创建文件
- 路径：
_knowledge_base/research-<主题>-<YYYYMMDD>.md - 初始内容包含：调研目标、关键问题、预期输出
# [主题] 调研笔记
调研日期：YYYY-MM-DD
调研目标：[一句话说明]
## 关键问题
1. [问题1]
2. [问题2]
3. [问题3]
## 发现
（调研中逐步填充）
## 来源列表
（每次搜索后追加）Step 2: 搜索并增量保存
- 每次 WebSearch 后，立即将发现追加到文件
- 每条发现附上来源 URL 和日期
- 遵循信息源优先级（见 SHARED-RULES.md）
Step 3: 阶段摘要
- 每完成3次搜索，在文件中保存一次「阶段摘要」
- 格式：
### 阶段摘要 (第N轮)+ 当前关键发现
Step 4: 最终简报
调研结束时，整理文件为结构化简报：
## 调研结论
### 关键事实
1. [事实1]（来源：URL）
2. [事实2]（来源：URL）
### 来源列表
| 来源 | URL | 发布日期 | 可信度 |
|------|-----|---------|--------|
| ... | ... | ... | 高/中/低 |
### 待确认问题
- [还需要进一步验证的点]
### 写作建议
- [基于调研结果，对后续写作的建议]关键原则
- 先建文件再搜索：确保第一次搜索结果就被保存
- 增量保存不等到最后：每次搜索后立即追加
- 调研和写作分离：本 Skill 只做调研，不开始写草稿
- 标注可信度：区分一手信息（官方）和二手信息（媒体/社区）
- 忽略过时信息源：知乎/百度（2025年前）、营销软文
与其他 Skill 的关系
- 调研完成后，用户可触发 /选题生成 来确定写作方向
- 调研文件将作为后续写作的输入素材
- 如果调研中发现的信息适合长期留存，保存到对应的 _knowledge_base 分类目录
输出位置
- 调研笔记：
_knowledge_base/research-<主题>-<YYYYMMDD>.md - 长期知识：
_knowledge_base/<分类>/<主题>-<YYYYMM>.md
最后更新: 2026-02-06
---
花叔出品 | AI Native Coder · 独立开发者
公众号「花叔」| 30万+粉丝 | AI工具与效率提升
代表作：小猫补光灯（AppStore付费榜Top1）·《一本书玩转DeepSeek》
Related skills
Grill MeHave an agent relentlessly pressure-test their plan or design until every decision branch is resolved and shared understanding is reached.691k196kGrill With DocsPressure-test architecture decisions against their live codebase and living documentation before writing production code.585k196k
Lark CalendarLet their coding agents read, create, and manage events inside Lark (Feishu) Calendar directly from natural language instructions.493kLark ImLet their coding agent send instant notifications, post updates, and receive commands directly inside Lark (Feishu) workspaces.493kLark ContactLet their AI coding agents read and write contacts, send messages, and manage user data inside Lark (Feishu).493kLark AttendanceAutomatically fetch and report daily attendance data from Lark (Feishu) into their own tools or agents.493kFAQ
Where does huashu-research save findings?
huashu-research creates files at `_knowledge_base/research-<主题>-<YYYYMMDD>.md` before the first search, then appends discoveries and source URLs after each WebSearch round.
When does huashu-research write stage summaries?
huashu-research saves a stage summary labeled `### 阶段摘要 (第N轮)` after every 3 completed searches, capturing key findings before the final briefing.