github-standup-agent 0.3.0

AI-powered daily standup summaries from GitHub activity using OpenAI Agents SDK

GitHub Standup Agent

AI-powered daily standup summaries from your GitHub activity, built with the OpenAI Agents SDK.

Features

• Automatic GitHub Activity Collection: Pulls your PRs, issues, commits, and code reviews using the gh CLI
• Slack Integration: Read team standups for context and publish your standup directly to Slack threads
• AI-Powered Summarization: Creates concise, well-formatted standup summaries
• Interactive Chat Mode: Refine your standup through conversation ("make it shorter", "ignore the docs PR")
• Session Persistence: Resume chat sessions later, use named sessions for recurring standups
• Style Customization: Define your team's standup format with style.md and example standups
• Historical Context: References past standups to maintain continuity and avoid repetition
• Multiple Output Options: Print to terminal, copy to clipboard, or publish to Slack
• Fully Local: Your data never leaves your machine (except for OpenAI/Slack API calls)

Architecture

This project showcases many features of the OpenAI Agents SDK:

• Multi-Agent Workflow: Coordinator → Data Gatherer → Summarizer
• Function Tools: @function_tool decorated functions for GitHub CLI operations
• Handoffs: Agents delegate to specialized sub-agents
• Guardrails: Input validation and PII detection
• Hooks: Logging and observability
• Sessions: SQLite-based history for past standups

Installation

Prerequisites

1. GitHub CLI: Install and authenticate with gh

   # macOS
   brew install gh
   
   # Then authenticate
   gh auth login
   

2. OpenAI API Key: Get one from OpenAI

Install from PyPI

pip install github-standup-agent

Or with uv:

uv pip install github-standup-agent

Install from Source

git clone https://github.com/andymaguire/github-standup-agent
cd github-standup-agent
uv pip install -e ".[dev]"

Usage

Set your OpenAI API key

export OPENAI_API_KEY="sk-..."

Generate a Standup (One-Shot)

# Generate standup for today
standup generate

# Look back 3 days
standup generate --days 3

# Copy directly to clipboard
standup generate --output clipboard

# Include context from recent standups
standup generate --with-history

Interactive Chat Mode

Start an interactive session to refine your standup:

standup chat

Example session:

> generate my standup

**Yesterday:**
- Merged PR #123: Add user authentication
- Reviewed PR #456: Fix login bug
- Closed issue #789: Update documentation

**Today:**
- Working on PR #124: Add OAuth support (in progress)
- Assigned to issue #790: Performance optimization

**Blockers:**
- Waiting for API access from platform team

> make it less wordy

**Yesterday:** Merged auth PR, reviewed login fix, updated docs
**Today:** OAuth support PR, performance optimization
**Blockers:** Waiting on platform team API access

> ignore the docs, focus on auth work

**Yesterday:** Merged user authentication PR #123
**Today:** Continuing OAuth support (PR #124)
**Blockers:** Waiting on platform team API access

> perfect, copy to clipboard
✅ Copied to clipboard!

> publish to slack
✅ Posted standup to #standups

View History

# List recent standups
standup history --list

# View a specific date
standup history --date 2025-01-14

# Clear history
standup history --clear

Chat Sessions

Chat sessions are automatically saved and can be resumed:

# Start a new session (auto-named by date)
standup chat

# Resume the last session
standup chat --resume

# Use a named session
standup chat --session weekly

# List recent sessions
standup sessions --list

# Clear all sessions
standup sessions --clear

Style Customization

Customize how your standups are generated to match your team's format:

# Set quick style instructions
standup config --set-style "Use 'Did:' and 'Will Do:' sections. Be concise."

# Create and edit a detailed style guide
standup config --init-style
standup config --edit-style

Example Standups (Few-Shot Prompting)

Provide real examples of standups you like. The AI uses these to match your preferred tone and format:

# Create and edit examples file
standup config --init-examples
standup config --edit-examples

See style.example.md and examples.example.md for templates.

File locations: The CLI checks the current directory first, then ~/.config/standup-agent/. This lets you keep project-specific styles in your repo (gitignored).

Slack Integration

Optionally connect to Slack to read team standups and publish your own:

# Set your Slack channel
standup config --set-slack-channel standups

# Set your Slack bot token (keep this secure)
export STANDUP_SLACK_BOT_TOKEN="xoxb-..."

Required Slack Bot Permissions:

• channels:history - Read messages in public channels
• channels:read - View basic channel info
• chat:write - Post messages

In chat mode, you can publish directly to Slack:

> generate my standup
[standup generated]

> publish to slack
Preview: Your standup will be posted to #standups...
Please confirm by saying "yes" or "publish it"

> yes
✅ Posted standup to #standups

The agent reads existing standup threads (posts starting with "Standup") and publishes your standup as a reply to the most recent thread.

Configuration

# Show current config
standup config --show

# Set GitHub username (auto-detected by default)
standup config --set-github-user myusername

# Set model for summarization
standup config --set-model gpt-5.2

Configuration Options

Environment Variable	Description	Default
OPENAI_API_KEY	OpenAI API key (required)	-
STANDUP_GITHUB_USER	GitHub username	Auto-detected
STANDUP_COORDINATOR_MODEL	Model for coordinator agent	gpt-5.2
STANDUP_DATA_GATHERER_MODEL	Model for data gathering	gpt-5.2
STANDUP_SUMMARIZER_MODEL	Model for summarization	gpt-5.2
STANDUP_SLACK_BOT_TOKEN	Slack bot token for integration	-
STANDUP_SLACK_CHANNEL	Default Slack channel (or use config)	-

Config file location: ~/.config/standup-agent/config.json

Development

# Clone the repo
git clone https://github.com/andymaguire/github-standup-agent
cd github-standup-agent

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

# Run tests
pytest

# Run linting
ruff check src/

# Run type checking
mypy src/

How It Works

┌─────────────────────────────┐
│    Coordinator Agent        │
│  (Orchestrates workflow)    │
└────────────┬────────────────┘
             │ handoffs
     ┌───────┴───────┐
     ▼               ▼
┌──────────────┐ ┌──────────────┐
│ Data Gatherer│ │  Summarizer  │
│  (gpt-5.2)   │ │  (gpt-5.2)   │
└──────────────┘ └──────────────┘
       │                │
       ▼                ▼
  GitHub CLI        Past Standups
  (gh commands)     (SQLite DB)

1. Coordinator receives your request and routes it appropriately
2. Data Gatherer uses gh CLI tools to collect PRs, issues, commits, reviews
3. Summarizer creates a formatted standup, referencing past standups for context
4. You can refine through chat, copy to clipboard, or save to history

License

MIT License - see LICENSE for details.

Contributing

Contributions welcome! Please open an issue or PR.

Acknowledgments

• Built with the OpenAI Agents SDK
• Uses the GitHub CLI for data gathering
• CLI powered by Typer