Build a No‑Code Personal Work Log with Claude Code, Obsidian, and Markdown in 30 Minutes

Problem

Developers often perform fragmented tasks throughout the day and forget them by Friday, making weekly report writing time‑consuming and incomplete.

Solution Architecture

An automated personal work‑log system stores all data as plain Markdown files and uses Claude Code skills as the processing engine. Obsidian can be used to visualize the Markdown vault.

Directory Layout

work-log/
├── CLAUDE.md               ← system rules (loaded by AI)
├── .claude/skills/
│   └── work-log/
│       └── skill.md        ← core engine (classification + processing)
└── kb/                     ← knowledge base (Markdown files)
    ├── inbox/               ← quick daily entries
    ├── daily/               ← formal daily logs (year/month)
    ├── weekly/              ← auto‑generated weekly reports
    ├── monthly/             ← auto‑generated monthly reports
    └── tags/                ← tag index for project/skill lookup

Environment Setup (≈5 minutes)

Install Claude Code: npm install -g @anthropic-ai/claude-code Create project directory: mkdir work-log-system && cd work-log-system Create required sub‑directories:

mkdir -p kb/{inbox,daily,weekly,monthly,tags,templates,plans}
mkdir -p .claude/skills/work-log

System Rules (CLAUDE.md)

# CLAUDE.md — 工作日志知识库系统

知识库根目录: `kb/`

## 分类体系
| 类别 | 类型标识 | 识别关键词 |
|------|----------|-----------|
| 需求开发 | dev | 做了/完成/开发/实现/编写 |
| 上线发布 | release | 上线/发布/部署/灰度 |
| 需求评审 | review | 评审/过需求/需求评审 |
| 技术设计 | design | 设计/方案/架构/技术评审 |
| Bug 修复 | bugfix | 修/修复/排查/解决 |
| 数据修复 | datafix | 数据修复/数据修正 |
| 面试 | interview | 面试/面了/候选人 |
| 学习成长 | learn | 学习/看了/研究/了解 |
| 会议 | meeting | 开会/会议/同步/对齐 |

## 文件规则
### 记录工作时
1. 解析输入，自动识别类别
2. 写入 `kb/inbox/YYYY-MM-DD.md`
3. 同时写入 `kb/daily/YYYY/MM/YYYY-MM-DD.md`
4. 更新 `kb/tags/index.md`（标签索引）

### 生成周报时
1. 扫描本周 daily 日志
2. 按类型分类汇总
3. 生成中文周报 → `kb/weekly/WXX.md`

### 生成月报时
1. 汇总本月所有日志
2. 生成中文月报 → `kb/monthly/YYYY-MM.md`

Skill Engine (skill.md)

---
name: work-log
version: 1.0.0
description: Record daily work, auto‑classify, generate weekly reports.
allowed-tools: [Read, Write, Edit, Bash, Glob, Grep]
---

# /log — 工作日志记录

## 分类规则
dev → 需求开发
release → 上线发布
review → 需求评审
design → 技术设计
bugfix → Bug 修复
datafix → 数据修复
interview → 面试
learn → 学习成长
meeting → 会议

## 处理流程
1. 解析输入（自然语言或 /log 命令）
2. 获取当前日期、周号、月份
3. 写入 inbox + daily 文件
4. 更新标签索引

Tip: The full version can handle fuzzy input, boundary cases, synonym merging, and historical linking.

Templates

Tag Index (kb/tags/index.md)

# 标签索引

## `#dev`（需求开发）
## `#release`（上线发布）
## `#review`（需求评审）
## `#bugfix`（Bug 修复）
## `#datafix`（数据修复）
## `#learn`（学习成长）
## `#meeting`（会议）
## `#interview`（面试）

> 每次记录新条目时自动更新

Weekly Report Template (kb/templates/weekly-summary.md)

# 工作周报 - {{week}}

## 本周关键摘要
**本周亮点**：...
**需关注**：...
**下周重点**：...

## 一、需求开发与上线
### 1.1 开发中需求
- 需求名称：具体工作（来源：日期）
### 1.2 已上线需求
- 《需求名称》：上线说明（来源：日期）

## 二、需求评审

## 三、Bug 修复

## 四、数据修复

## 五、学习成长

## 六、下周计划

## 七、风险与问题

Using the System

Start Claude Code:

claude

Record a Work Item

我今天完成了用户管理模块的开发，完成了登录和注册的 API 接口

Claude automatically:

Classifies the entry as dev (需求开发).

Creates kb/inbox/2026-04-29.md.

Creates kb/daily/2026/04/2026-04-29.md.

Updates kb/tags/index.md under #dev.

Record Additional Items

下午评审了订单模块的需求

修了登录接口的 token 过期问题

晚上学习了 React Server Components

The three entries are classified as review , bugfix , and learn respectively and appended to the same daily file.

Generate a Weekly Report

生成周报

Claude produces a formatted Chinese weekly report in about three seconds:

# 工作周报 - 2026-W18

## 本周关键摘要
**本周亮点**：用户管理模块 API 开发完成
**需关注**：token 过期问题需跟进
**下周重点**：推进录单 8.1.0 联调

## 一、需求开发与上线
- **用户管理模块**：完成登录和注册 API 开发（来源：2026-04-29）

## 二、需求评审
- **订单模块需求评审**：提出 3 个问题待确认（来源：2026-04-29）

## 三、Bug 修复
- **登录接口 token 过期问题**：修复 token 刷新逻辑（来源：2026-04-29）

## 四、学习成长
- **React Server Components**：阅读官方文档，了解 SSR 优化（来源：2026-04-29）

Advanced Features

Automatic Tag Extraction

If an entry contains a pattern like 【项目名】, the system extracts tags automatically.

完成了产品b的录取需求，订单中心对接联调完成

Extracted tags: #录取 、 #产品 、 #订单中心.

Sub‑type Labels

Use sub‑type syntax to distinguish front‑end, back‑end, or API work:

/log dev:frontend 完成登录页面开发
/log dev:backend 实现用户认证接口
/log dev:api 联调录单接口

Generated reports display sub‑type tags such as [前端] 、 [后端] 、 [接口].

Synonym Merging

Create kb/tags/synonyms.md to map synonyms to a primary tag.

#synonyms
#前端 => #frontend
#bug => #缺陷

When a synonym is used, it is automatically converted to the primary tag.

Obsidian Visualization

Calendar view : browse entries by date.

Tag filtering : click a tag (e.g., #产品) to see related records.

Graph view : visualize tag relationships.

Full‑text search : locate entries instantly.

Extension Ideas

Daily reminder hook: Claude prompts “今天的工作记录了吗？” before leaving work.

Push generated reports to DingTalk or Feishu via API.

Collaborative mode: store the project in a Git repository; each team member records entries and weekly reports can be generated per person.

Monthly review: automatically tally completed features, bug fixes, learning time, and project effort distribution when generating a monthly report.

Core Takeaway

The system relies solely on local Markdown files and Claude Code skills; no database or server is required. Recording work with natural‑language input and automatic classification reduces the time spent on weekly reporting and provides structured data for monthly retrospectives.