agent-daily-review（Agent 每日回顾）

Agent Daily Review

Overview

The Daily Review skill helps agents conduct systematic review and reflection at the end of the day, transforming fragmented daily records into structured growth accumulation.

Core Capabilities:

1. Scan Records - Automatically scan today's memory files, artifacts, and MEMORY.md entries
2. Categorize Activities - Classify activities into: Completed, In Progress, Issues/Blockers, Learning/Growth, Others
3. Reflect and Analyze - Calculate productivity score, identify highlights and challenges, generate improvement suggestions
4. Generate Report - Output structured review report and archive to long-term memory

Use Cases:

• User says "Do my daily review for today"
• User says "Summarize today"
• Cron scheduled task triggers (e.g., daily at 22:00)
• User wants to review work/learning status for a specific day

Workflow

1. Scan Today's Records

Execute scripts/daily_review.py to scan the following:

• memory/YYYY-MM-DD.md - Today's journal entries
• MEMORY.md - Today's entries in long-term memory
• workspace/*.md - Artifact files generated today

2. Categorize Activities

Automatically identify and categorize:

• Completed - Contains keywords like "completed," "done," "resolved," ✅
• In Progress - Contains keywords like "in progress," "working on," 🔄
• Issues/Blockers - Contains keywords like "issue," "blocked," "bug," ❌
• Learning/Growth - Contains keywords like "learned," "researched," "understood"
• Meetings/Communication - Contains keywords like "meeting," "discussed," "sync"

3. Reflect and Analyze

Perform intelligent analysis based on categorization results:

• Productivity Score - Calculate based on record count and artifact count (0-100)
• Today's Highlights - Identify completed important tasks and decisions
• Challenges Encountered - Summarize issues and pending items
• Improvement Suggestions - Generate personalized recommendations based on data

4. Generate Report

Output structured report containing:

• Today's Overview (statistics)
• Completed Tasks List
• In Progress Tasks List
• Issues/Blockers
• Learning/Growth Records
• Highlights Summary
• Reflection and Suggestions
• Tomorrow's Plan Framework

5. Archive to Memory

• Save review report to reviews/review_YYYY-MM-DD.md
• Append review summary to MEMORY.md

Usage

Manual Execution

# Execute today's review
python scripts/daily_review.py

# Specify working directory
python scripts/daily_review.py -w /path/to/workspace

# Specify output file
python scripts/daily_review.py -o /path/to/output.md

# Review specific date
python scripts/daily_review.py -d 2024-01-15

# Do not save to MEMORY.md
python scripts/daily_review.py --no-memory

Use as Module

from scripts.daily_review import DailyReview

review = DailyReview("/path/to/workspace")
report = review.run(save_to_memory=True)
print(report)

Cron Auto-Trigger

Set up automatic daily review at 22:00:

# Add scheduled task using openclaw cron
openclaw cron add --name "daily-review" \
  --schedule "0 22 * * *" \
  --command "python ~/.qclaw/skills/daily-review/scripts/daily_review.py"

Or using cron tool:

{
  "name": "daily-review",
  "schedule": {"kind": "cron", "expr": "0 22 * * *", "tz": "Asia/Shanghai"},
  "payload": {
    "kind": "agentTurn",
    "message": "Please perform today's review using the daily-review skill, scanning today's records and generating a review report."
  },
  "sessionTarget": "isolated"
}

Report Format

Review reports use Markdown format with the following sections:

# Daily Review Report - YYYY-MM-DD

## 📊 Today's Overview
- Date, Record Count, Artifact Count, Productivity Score

## ✅ Completed
- Task List

## 🔄 In Progress
- Pending List

## ⚠️ Issues/Blockers
- Issue List

## 📚 Learning/Growth
- Learning Records

## 🎯 Today's Highlights
- Highlights Summary

## 💭 Reflection and Suggestions
- Improvement Suggestions

## 📝 Tomorrow's Plan
- Plan Framework

Directory Structure

workspace/
├── memory/
│   └── 2024-01-15.md          # Today's journal entries
├── reviews/
│   └── review_2024-01-15.md   # Review report
├── MEMORY.md                   # Long-term memory (review summary appended here)
└── *.md                        # Artifacts generated today

Tips

1. Cultivate Journaling Habit - Record timestamped entries in memory/YYYY-MM-DD.md daily for better review results
2. Use Keywords - Use keywords like "completed," "learning," "encountered issue" when journaling to facilitate auto-categorization
3. Periodic Review - Review weekly/monthly review reports to discover growth trajectory
4. Integrate with Cron - Set up automatic review to ensure daily reflection is never missed

Resources

• scripts/daily_review.py - Core review script
• references/framework.md - Detailed review framework explanation (optional reading)