Automatically capture your daily work with intelligent activity detection
Never miss documenting your work again. daily-log automatically detects what you've been working on (experiments, features, bugs, learning) and generates structured daily logs with appropriate questions for each activity type.
- 🤖 Auto-Generation Mode - No more questions! AI automatically generates detailed logs by analyzing your git commits, diffs, and code
- 🌍 Automatic Language Detection - Write in your language, get logs in your language (English, Korean, Japanese)
- ⚡ Interactive Mode - Add
--interactiveflag when you want to answer questions manually
- 🎯 No setup required - Works for anyone, any role
- 🔍 Automatic activity detection - Experiments, Features, Bug Fixes, Refactoring, Learning, Documentation, Testing
- 🤖 Auto-generation - AI analyzes code and generates comprehensive logs (no questions!)
- 🌍 Multi-language - Automatic language detection from your request
- 📝 Never misses work - Tracks last log timestamp, even with multi-day gaps
- 💡 Activity-specific questions - Each activity type gets relevant questions (in interactive mode)
- ✅ Quality standards - Enforces detailed answers (80+ words per item)
- 🚀 Multi-source detection - Git commits + uncommitted changes + untracked files
One-line install:
curl -fsSL https://raw.githubusercontent.com/wlsdnen/daily-log-skill/main/install.sh | bashOr manual install:
mkdir -p ~/.claude/skills/daily-log
curl -o ~/.claude/skills/daily-log/SKILL.md \
https://raw.githubusercontent.com/wlsdnen/daily-log-skill/main/SKILL.mdOr clone and install:
git clone https://github.com/wlsdnen/daily-log-skill.git
cd daily-log-skill
./install.shRecommended: Use Git Bash (comes with Git for Windows)
# 1. Install Git for Windows (if not already installed)
# Download from: https://git-scm.com/download/win
# 2. Open Git Bash and run the installation command
curl -fsSL https://raw.githubusercontent.com/wlsdnen/daily-log-skill/main/install.sh | bash
# 3. (Optional) Install jq for better performance
# Download from: https://github.com/stedolan/jq/releases/latest/download/jq-win64.exe
# Place in: C:\Program Files\Git\usr\bin\jq.exeAlternative: Use WSL (Windows Subsystem for Linux)
- Install WSL:
wsl --installin PowerShell (admin) - Follow Linux installation steps inside WSL
Note: The skill uses bash scripts and works in Git Bash or WSL. Native PowerShell/CMD is not supported.
In your project directory:
"Create daily log" # Auto-generates in English
"오늘 작업 기록해줘" # Auto-generates in Korean
"Create daily log --interactive" # Ask questions mode
The skill will:
- Detect your language from the request
- Analyze all changes since your last log
- Automatically generate comprehensive answers by reading commits and code
- Create a log in
./daily-notes/YYYY-MM-DD.md
# 2026-02-05 Daily Log
_Period: 2026-02-03 09:00 ~ 2026-02-05 18:30 (2 days)_
## Experiments (detected from: src/experiment.ts)
### Parallel Processing Performance Test
- **Hypothesis**: Testing if chunked parallel processing achieves 2x+ improvement
- **Method**: Implemented parallelProcess() with 100-item chunks using Promise.all
- **Results**: 2x improvement for 10k items (400ms → 200ms), but memory spikes at 100k items
- **Learnings**: Promise.all is concurrent, not parallel. Need Worker threads for true parallelism
- **Next**: Implement Worker thread pool with configurable max concurrency
## Bug Fixes (detected from: fix: typo commit)
### Documentation Typo Fix
- **Problem**: Incorrect method name in API documentation
- **Root Cause**: Copy-paste error from old version
- **Solution**: Updated all references to correct method name
- **Prevention**: Added documentation review checklist
- **Verification**: Built docs, checked all linksSee full example for complete output.
- Git commits capture only 10-30% of daily work
- Most important work (experiments, in-progress changes) is uncommitted
- Manual logging is tedious and easy to forget
- Different roles need different documentation styles
- Automatic detection - Analyzes git history, file changes, and commit messages
- Dynamic sections - Only creates sections for activities you actually did
- Smart questions - Asks relevant questions based on activity type
- Never misses work - State tracking ensures continuity even with gaps
- Migration Guide - Upgrading from v1.x (profile-based)
- Example Log - See what generated logs look like
- SKILL.md - Complete skill documentation
Breaking changes from v1.x:
- ✅ Removed profile system (Researcher/Developer/Learner/Hybrid)
- ✅ Added dynamic activity detection
- ✅ Added state tracking for continuity
- ✅ Now works for anyone without setup
Migration: Just install v2.1 - it works automatically, no configuration needed.
- State Tracking - Remembers last log timestamp in
~/.claude/daily-log-state.json - Multi-Source Detection - Checks git log, staged/unstaged changes, untracked files (7 sources)
- Activity Classification - Matches changes against patterns (experiments/, feat:, fix:, etc.)
- Dynamic Generation - Creates sections only for detected activities
- Quality Enforcement - Ensures detailed answers with follow-up questions
| Activity | Detection Pattern | Questions Asked |
|---|---|---|
| Experiments | experiment*.*, experiments/ paths |
Hypothesis, Method, Results, Learnings, Next |
| Features | feat:, add: commits, new files |
What, How, Why (alternatives), Effect, Issues |
| Bug Fixes | fix:, bug: commits |
Problem, Root Cause, Solution, Prevention, Verification |
| Refactoring | refactor:, clean commits |
Reason, Changes, Improvements, Metrics |
| Learning | notes/, tutorials/ paths, .md files |
Topic, Difficulty, Breakthrough, Application |
| Documentation | docs/, README files |
Purpose, Audience, Improvements |
| Testing | test/, *.test.* files |
Coverage, Findings, Issues |
- Minimum 80-100 words per item
- Specific details (not generic "good"/"fast")
- Quantitative metrics when possible
- Alternatives considered
- Context for future review
The skill enforces these standards with follow-up questions if answers are too brief.
- Claude Code CLI
- Git Bash (Windows), bash (macOS/Linux), or WSL
- Git repository (optional but recommended)
jq(for JSON parsing) - optional, skill has fallback if missing
- macOS/Linux: Works out of the box
- Windows: Requires Git Bash (included with Git for Windows) or WSL
- jq installation: Optional but recommended for better performance
- macOS:
brew install jq - Linux:
apt install jqoryum install jq - Windows (Git Bash): Download from jq releases
- macOS:
Issues and PRs welcome! This is v2.1 - let me know if you find bugs or have feature requests.
MIT License - see LICENSE file
Built with Claude Code's skill system. Tested extensively with 6 test scenarios achieving 100% pass rate.
Version: 2.1.0 Status: Production Ready ✅ Last Updated: 2026-02-05