🐶 RepoWatchDog

RepoWatchDog is a lightweight GitHub Actions-powered watchdog that automatically monitors GitHub repositories weekly and delivers a formatted summary report as a GitHub Issue.

It tracks:

• 🤖 AI Summary – a concise Traditional-Chinese summary of the week's most important changes (powered by GitHub Models / gpt-4o-mini)
• 🔀 Merged PRs – pull requests merged this week
• 🚀 New Releases – what changed, release notes included
• 🐛 Important Issues – newly opened issues filtered by configurable labels
• 📊 Commit Activity – total commit count and top-3 contributors

Default target repo: github/copilot-sdk (GitHub Copilot SDK) — fully configurable.

---

How It Works

1. A scheduled GitHub Actions workflow runs every Monday at 09:00 UTC.
2. The Python script (scripts/watch_dog.py) queries the GitHub API for new releases, merged PRs, issues, and commit activity since the last check.
3. An AI summary is generated (via GitHub Models) describing the week's key changes.
4. A Markdown summary report is generated.
5. The report is posted as a GitHub Issue (labelled weekly-report) in this repository.
6. The last-check timestamp is committed back to state/last_check.json so the next run only covers the new period.

---

Setup

1. Fork / clone this repository

git clone https://github.com/hondayaya123/RepoWatchDog.git
cd RepoWatchDog

2. Configure which repos to watch

Edit config.json:

{
  "watch_repos": [
    {
      "owner": "github",
      "repo": "copilot-sdk",
      "description": "GitHub Copilot SDK"
    }
  ],
  "report_repo": {
    "owner": "",
    "repo": ""
  },
  "lookback_days": 7,
  "important_labels": ["bug", "enhancement", "breaking change", "priority/high"],
  "ai_summary": true
}

• watch_repos – list of repos to monitor (add as many as you like)
• report_repo – where to create the summary issue (leave empty to use the current repo via REPORT_OWNER / REPORT_REPO env vars)
• lookback_days – maximum look-back window (default: 7)
• important_labels – only issues with at least one of these labels appear in the report; set to [] to show all issues
• ai_summary – set to true to prepend an AI-generated Traditional-Chinese summary (requires GITHUB_TOKEN with GitHub Models access)

3. Enable GitHub Actions

The workflow .github/workflows/weekly_report.yml uses the built-in GITHUB_TOKEN, so no extra secrets are needed for basic usage.

Make sure Issues are enabled on your repository (Settings → Features → Issues ✅).

4. (Optional) Run manually

Go to Actions → Weekly Repo Summary Report → Run workflow and choose how many days to look back.

---

Local Development

pip install -r requirements.txt

# Set your token
export GITHUB_TOKEN=ghp_...
export REPORT_OWNER=your-github-username
export REPORT_REPO=RepoWatchDog

python scripts/watch_dog.py

Run tests

pip install pytest
pytest tests/

---

Project Structure

RepoWatchDog/
├── .github/
│   └── workflows/
│       └── weekly_report.yml   # Scheduled workflow
├── scripts/
│   └── watch_dog.py            # Core logic
├── state/
│   └── last_check.json         # Persisted last-check timestamp
├── tests/
│   └── test_watch_dog.py       # Unit tests
├── config.json                 # Watched repos configuration
├── requirements.txt
└── README.md

---

New Features

🔀 Merged PRs Tracking

The report now includes a dedicated Merged PRs section per repository, showing PR number, title, link, merge date, and author.

🐛 Issue Importance Filtering

Set important_labels in config.json to only surface issues that matter. For example:

"important_labels": ["bug", "enhancement", "breaking change", "priority/high"]

Set to [] to show all issues (original behaviour).

🤖 AI Summary (GitHub Models)

When "ai_summary": true, the script calls the GitHub Models API (gpt-4o-mini via https://models.inference.ai.azure.com) and prepends a Traditional-Chinese summary (≤ 150 characters) describing the week's most important technical changes. The same GITHUB_TOKEN is used as the API key. If the call fails the report is still generated normally.

📊 Commit Activity Statistics

Each repo section now includes a 開發活躍度 block showing:

• Total commit count for the week
• Top-3 contributors (name + commit count)

---

Example Report

# 📦 RepoWatchDog Weekly Summary

**Report generated:** 2024-06-17 09:00 UTC
**Period covered:** 2024-06-10 09:00 UTC → 2024-06-17 09:00 UTC

## 🤖 AI 本週摘要
本週 copilot-sdk 最重要的變化是新版本 v1.2.0 的發布，並修復了多項關鍵問題。

---

## 🔍 github/copilot-sdk

### 🔀 Merged PRs (5)
- [#927 ephemeral events 改版](https://github.com/...) – 2024-06-12 by @user1

### 🚀 New Releases (1)
#### [v1.2.0](https://github.com/...) `v1.2.0` – 2024-06-12
> - Added new feature
> - Fixed a bug

### 🐛 重要 Issues (3)
- 🟢 [#99 Support for new model](https://github.com/...) [`enhancement`] – 2024-06-11
- 🟢 [#100 Crash on startup](https://github.com/...) [`bug`] – 2024-06-13

### 📊 開發活躍度
- 本週 commits：23
- 前三貢獻者：user1(10), user2(8), user3(5)