A simple CLI tool that uses AI agents to generate daily work summaries from your GitHub activity.

Analyzes your GitHub activity and generates daily summaries like this:

Yesterday:
- Merged PR #142: Fix authentication bug in user login flow
- Reviewed PR #138: Add support for OAuth2 integration
- Opened issue #145: Memory leak in background task processor

Today:
- Continue work on OAuth2 integration testing
- Address memory leak issue in background processor
- Review pending PRs from team members

Uses a simple agent powered by Mozilla AI's any-agent framework with any-llm for LLM provider abstraction. The agent connects to GitHub via Model Context Protocol (MCP) to analyze your activity and generate summaries.

• Python 3.11 or newer
• A GitHub Personal Access Token
• API key for your chosen LLM provider (OpenAI, Anthropic, etc.)

pip install dailysum

git clone https://github.com/njbrake/dailysum.git
cd dailysum
pip install -e .

Run the initialization command and follow the prompts:

dailysum init

This will:

• Prompt for your GitHub token
• Let you choose your preferred LLM model
• Optionally set your company name
• Save configuration to ~/.config/dailysum/config.toml

Set these environment variables:

export GITHUB_TOKEN="ghp_your_github_token_here"
export MODEL_ID="openai/gpt-4o-mini"  # Optional, defaults to gpt-4o-mini
export COMPANY="Your Company Name"     # Optional

Then run with the --use-env flag:

dailysum generate --use-env

1. Go to GitHub Settings > Developer settings > Personal access tokens
2. Click "Generate new token (classic)"
3. Select scopes: repo, read:user, read:org, notifications
4. Copy the generated token

• OpenAI: openai/gpt-4o, openai/gpt-4o-mini, openai/gpt-3.5-turbo
• Anthropic: anthropic/claude-3-5-sonnet-20241022, anthropic/claude-3-haiku-20240307
• Mistral: mistral/mistral-large-latest, mistral/mistral-small-latest
• Google: google/gemini-1.5-pro, google/gemini-1.5-flash

See any-llm providers for the complete list.

dailysum generate

dailysum config

dailysum generate --use-env

# Initialize with custom location
dailysum init --config-path /path/to/my/config.toml

# Generate using custom location
dailysum generate --config-path /path/to/my/config.toml

You can easily switch between models by updating your config:

# Use a faster, cheaper model
dailysum init --model-id "openai/gpt-4o-mini"

# Use a more powerful model for complex summaries
dailysum init --model-id "anthropic/claude-3-5-sonnet-20241022"

git clone https://github.com/njbrake/dailysum.git
cd dailysum

# Install with development dependencies
pip install -e ".[dev]"

# Set up pre-commit hooks
pre-commit install

pytest

# Format and lint
ruff format .
ruff check . --fix

# Type checking
mypy src/

Contributions are welcome. Areas of interest:

• New LLM provider support
• Summary templates and prompts
• Additional GitHub data sources
• CLI improvements
• Tests and documentation

See Contributing Guide for details.

📋 Your Daily Summary
┌─────────────────────────────────────────────────────────────┐
│ Yesterday:                                                  │
│ - Merged PR #234: Implement user authentication system     │
│ - Reviewed PR #231: Add Docker configuration               │
│ - Fixed critical bug in payment processing (Issue #189)    │
│ - Updated documentation for API endpoints                  │
│                                                             │
│ Today:                                                      │
│ - Complete integration tests for authentication system     │
│ - Review pending PRs from team members                     │
│ - Start work on user dashboard redesign                    │
│ - Investigate performance issues in search functionality   │
└─────────────────────────────────────────────────────────────┘

Make sure you've either:

• Run dailysum init to set up a config file, or
• Set the GITHUB_TOKEN environment variable

Install the required dependencies:

pip install any-agent any-llm

If you hit GitHub API rate limits:

• Use a GitHub token (provides 5000 requests/hour vs 60 for unauthenticated)
• Consider running the tool less frequently
• The tool automatically respects rate limits and will wait if needed

MIT License - see LICENSE for details.

• any-llm - Unified LLM provider interface
• any-agent - Unified AI agent framework
• GitHub Copilot MCP - GitHub API access via Model Context Protocol
• Rich - Beautiful terminal output
• Click - Command-line interface framework

Built by Mozilla AI