GitHub Activity Digest

A GitHub Action that generates AI-powered weekly summaries of repository activity. Add to any workflow to get automated progress reports.

Quick Start

Add to your workflow:

- uses: abensur/github-activity-digest@v1
  with:
    mode: organization
    organization: my-org
    days: 7
  env:
    GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
    ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}

✨ Features

• 🎯 Flexible Sources - Organizations, users, topics, or custom lists
• 🤖 Multi-AI - Anthropic Claude or OpenAI GPT
• 📊 Comprehensive - PRs, commits, and code statistics
• ⚡ Fast - Parallel processing, caching, automatic retry
• 🌍 Multi-Language - Output in any language

Full Workflow Example

name: Weekly Activity Digest

on:
  schedule:
    - cron: '0 10 * * 5'  # Every Friday at 10:00 AM UTC
  workflow_dispatch:

permissions:
  contents: write

jobs:
  generate-summary:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Generate Weekly Summary
        id: digest
        uses: abensur/github-activity-digest@v1
        with:
          mode: organization
          organization: ${{ github.repository_owner }}
          days: 7
          ai-provider: anthropic
          language: English
          max-repos: 100
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
          ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}

      - name: Commit Summary
        run: |
          git config user.name "github-actions[bot]"
          git config user.email "github-actions[bot]@users.noreply.github.com"
          git add archive/
          git diff --staged --quiet || git commit -m "Weekly summary"
          git push

Action Inputs

Input	Description	Default
mode	Source mode: organization, user, topics, list	organization
organization	GitHub org name (organization mode)	-
user	GitHub username (user mode)	-
topics	Comma-separated topics (topics mode)	-
repositories	Comma-separated owner/repo (list mode)	-
days	Days to look back	7
ai-provider	anthropic or openai	anthropic
ai-model	AI model name	claude-sonnet-4-20250514
language	Output language	English
max-repos	Max repositories to process	500
exclude-repos	Comma-separated exclude patterns	-
include-repos	Comma-separated include patterns	-
only-public	Only public repos	false
only-private	Only private repos	false
archive-dir	Directory to save summaries	archive
output-format	Output format: markdown, json, or both	markdown
slack-webhook	Slack incoming webhook URL	-
discord-webhook	Discord webhook URL	-

Action Outputs

Output	Description
summary	The generated summary content
summary-file	Path to the saved summary file
json-file	Path to JSON file (if output-format includes json)
repos-processed	Number of repositories processed
active-repos	Number of repositories with activity

� Authentication

GitHub Token

The GITHUB_TOKEN is automatically provided by GitHub Actions. For organization access, create a Personal Access Token with repo and read:org scopes.

AI Provider Keys

Add to repository secrets (Settings → Secrets → Actions):

• Anthropic: Get API key →
• OpenAI: Get API key →

⚙️ Configuration

config.json

{
  "mode": "organization",
  "source": { "organization": "your-org" },
  "filters": {
    "excludeRepos": ["archived-", "test-"],
    "includeRepos": ["api-", "app-"],
    "maxRepos": 500
  },
  "period": { "days": 7 },
  "ai": {
    "provider": "anthropic",
    "model": "claude-sonnet-4-20250514",
    "maxTokens": 4000,
    "temperature": 1.0
  },
  "output": {
    "archiveDir": "archive",
    "language": "English"
  }
}

Available Modes

Mode	Description	CLI Example
organization	All repos from a GitHub org	--org facebook
user	All repos from any user	--user torvalds
topics	Repos by GitHub topics	--topics ai,ml
file	Custom list from file	--file repos.txt
list	Direct array in config	repositories: [...]

AI Providers

Anthropic Claude (default):

ANTHROPIC_API_KEY=sk-ant-xxx

OpenAI GPT:

OPENAI_API_KEY=sk-xxx

{ "ai": { "provider": "openai", "model": "gpt-4-turbo" } }

� CLI Usage (Development)

The action can also be run locally as a CLI tool:

# Install Bun
curl -fsSL https://bun.sh/install | bash

# Clone and install
git clone https://github.com/abensur/github-activity-digest
cd github-activity-digest
bun install

# Run
bun start -- --user torvalds --days 7
bun start -- --org vercel --days 7
bun start -- --topics nextjs,react

CLI Options

Options:
  --org <name>        Organization mode
  --user <name>       User mode
  --topics <list>     Topics mode (comma-separated)
  --file <path>       File mode (list of repos)
  --days <n>          Days to look back (default: 7)
  --no-cache          Force refresh, bypass cache
  --provider <name>   AI provider: anthropic or openai
  --model <name>      AI model name

📝 Output

Generated summaries are saved to archive/weekly-report-YYYY-MM-DD.md by default.

Example output:

# Weekly Activity Summary
Period: January 23 - January 30, 2025

## Overview
Significant development activity across 5 repositories with
12 merged pull requests and 47 direct commits.

## Repository Details

### api-service
- **12 files changed** | +450 / -120 lines
- Implemented JWT authentication system
- Added rate limiting middleware
- Fixed memory leak in WebSocket handler

### frontend-dashboard
- **8 files changed** | +280 / -90 lines
- Redesigned mobile navigation
- Performance optimizations (50% faster load time)
- Added dark mode support

🔧 Advanced Features

Smart Caching

Activity data is cached to .cache/ directory for 30 minutes. The cache persists across runs, making it ideal for CI/CD workflows.

For GitHub Actions, add caching to your workflow:

- uses: actions/cache@v4
  with:
    path: .cache
    key: github-activity-${{ github.run_id }}
    restore-keys: github-activity-

Force refresh: Use --no-cache to bypass cache and fetch the latest commits:

bun start -- --user torvalds --days 7 --no-cache

This is useful when you need to include commits that were just pushed.

Automatic Retry

Network requests automatically retry with exponential backoff (1s → 2s → 4s) on transient failures. Rate limit errors are handled gracefully.

Visual Progress

Real-time progress indicators show:

• Repository processing: ██████░░░░ 60% (12/20)
• Activity collection status
• Cache hit rate

Filters

Fine-tune which repositories to include:

{
  "filters": {
    "excludeRepos": ["archived-", "test-"],
    "includeRepos": ["api-", "service-"],
    "onlyPublic": false,
    "onlyPrivate": false,
    "maxRepos": 500
  }
}

📂 Project Structure

github-activity-digest/
├── index.ts                     # Main entry point (Citty CLI)
├── action.yml                   # GitHub Action definition
├── Dockerfile                   # Docker image for GitHub Action
├── entrypoint.sh                # Action entrypoint script
├── lib/
│   ├── config.ts                # Config loading with Zod validation
│   ├── ai.ts                    # Multi-provider AI client
│   ├── github.ts                # Repository fetching
│   ├── activity.ts              # Activity tracking & stats
│   ├── retry.ts                 # Exponential backoff logic
│   ├── cache.ts                 # File-based caching (30min TTL)
│   └── logger.ts                # Centralized logging (consola)
├── tests/
│   ├── config.test.ts           # Config & utility tests
│   ├── cache.test.ts            # Cache system tests
│   ├── ai.test.ts               # AI client tests
│   ├── retry.test.ts            # Retry logic tests
│   ├── github.test.ts           # GitHub integration tests
│   └── activity.test.ts         # Activity tracking tests
├── config.json                  # Main configuration
├── prompt-template.txt          # AI prompt customization
├── .env                         # API keys & secrets
├── .cache/                      # Cached activity data
└── archive/                     # Generated summaries

🔧 Development

# Run the tool
bun start

# Run tests
bun test

# Run tests in watch mode
bun test --watch

# Type check
bun run typecheck

# Watch mode (auto-restart on changes)
bun --watch index.ts

Local Action Testing

Test the GitHub Action locally using act:

# Install act
curl -s https://raw.githubusercontent.com/nektos/act/master/install.sh | sudo bash
# or with package manager:
# brew install act        # macOS
# yay -S act              # Arch Linux
# sudo dnf install act    # Fedora

# Create secrets file from .env
cp .env .secrets

# Run the workflow locally
act -j generate-summary --secret-file .secrets

# Run with specific event
act workflow_dispatch --secret-file .secrets

Building the Docker Image

# Build locally
docker build -t github-activity-digest .

# Test locally
docker run --rm \
  -e GITHUB_TOKEN="$GITHUB_TOKEN" \
  -e ANTHROPIC_API_KEY="$ANTHROPIC_API_KEY" \
  -e INPUT_MODE=user \
  -e INPUT_USER=torvalds \
  -e INPUT_DAYS=7 \
  github-activity-digest

🏗️ Architecture

Technology Stack

• Runtime: Bun (10-100x faster I/O than Node.js)
• Language: TypeScript with strict mode
• Validation: Zod schemas for runtime type safety
• CLI: Citty for declarative argument parsing
• Testing: Bun test with comprehensive coverage
• Logging: Consola for beautiful CLI output

Key Design Decisions

1. Zod Schemas: Eliminate manual validation with declarative schemas
2. Citty CLI: Auto-generated help, type validation, clean API
3. Modular Architecture: Each file has single responsibility
4. Bun Native APIs: Bun.file(), Bun.write() for performance
5. Smart Caching: 30min TTL reduces redundant API calls
6. Retry Logic: Exponential backoff handles transient failures

🌍 Multi-Language Support

Change the output language in your prompt template or config:

{
  "output": {
    "language": "Portuguese"
  }
}

The AI will generate summaries in your specified language.

💡 Use Cases

• Team Standups - Automated weekly progress reports
• Open Source Tracking - Monitor contributions across projects
• Organization Metrics - Bird's-eye view of all development
• Technology Monitoring - Follow specific tech ecosystems via topics
• Multi-Repo Projects - Unified view of related repositories

🔧 Troubleshooting

No repositories found:

• Verify GITHUB_TOKEN has required scopes (repo, read:org)
• Check organization/user name spelling
• Ensure you have access to the repositories

AI API errors:

• Confirm API key is valid and set in .env
• Check you have credits/quota available
• Verify model name matches provider (e.g., claude-sonnet-4-20250514)

Empty or sparse summaries:

• Increase time period: --days 30
• Verify repositories had activity in the period
• Check filters aren't excluding active repos

Slow performance:

• Check cache is working (look for "Using cached" messages)
• Reduce number of repositories being processed
• Verify network connectivity

Rate limit errors:

• GitHub: Increase GITHUB_TOKEN rate limits (use authenticated requests)
• AI providers: Wait for quota reset or upgrade plan
• Retry logic handles transient errors automatically

🤝 Contributing

Contributions welcome! Please:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing)
3. Add tests for new features
4. Ensure tests pass (bun test)
5. Submit a pull request

📄 License

MIT License - see LICENSE file for details.

🙏 Credits

Built with:

• Bun - Fast JavaScript runtime with native TypeScript
• @octokit/rest - GitHub REST API client
• @anthropic-ai/sdk - Claude AI integration
• consola - Elegant console logging

---

Made with ❤️ for better team visibility