Auto Standup Agent

A lightweight agent that pulls activity from GitHub and Notion, summarizes each team member’s progress using a language model, and sends a clean Slack update. Designed for teams who want async visibility without the overhead of manual standups.

What It Does

1. Fetches activity from GitHub (commits, PRs) and Notion (task status).
2. Maps activity to individual contributors using GitHub usernames and Slack IDs.
3. Uses a local or API-based language model to generate clear summaries (what was done, what’s in progress, blockers).
4. Formats and sends messages to Slack in a simple, readable format.

How It Works

The agent runs automatically via GitHub Actions (or can be run manually). Here’s the flow:

flowchart TD
    A[Scheduled Trigger - GitHub Action] --> B[Auth & Setup]
    
    B --> C[GitHub API]
    C --> C1[Pull commits in last 24h]
    C --> C2[Fetch PRs created/merged]
    C --> C3[Extract metadata - author, repo, message]

    B --> D[Notion  API]
    D --> D1[Fetch Task DB or Board]
    D --> D2[Parse status - In Progress, Blocked, Done]
    D --> D3[Match user activity with GitHub if possible]

    C3 --> E[Activity Aggregation Engine]
    D3 --> E

    E --> F[LLM Summarizer - GROQ API]
    F --> F1[Prompt engineering - What I did / Doing / Blockers]
    F1 --> G[Per-user summaries]

    G --> H[Final Stand-Up Builder]
    H --> H1[Format per-user messages]
    H --> H2[Sort + merge team-wide summary]

    H2 --> I[Output to Slack ]
    I --> I1[Use Webhook or Bot to send to channel]

    I1 --> Z[Done! Daily stand-up delivered]

Loading

---

Getting Started

1. Clone the repo and install dependencies

git clone https://github.com/emanalytic/auto-standup-agent.git
cd auto-standup-agent
pip install -r requirements.txt

---

Configuration Guide

A. Set your team’s GitHub and Slack info

Open the config.ini file and fill in your team details under [members]. This is how the agent links GitHub activity to Slack updates:

[members]
Eman_github = emanalytic
Eman_slack_id = U12345678

John_github = just-building
John_slack_id = U87654321

Add one line per user for both GitHub and Slack. IDs must match exactly.

---

B. Set your config

Also in config.ini, set the language model you’re using under [settings]:

[settings]
model = llama3-70b-8192

Update the slack channel you want to post the message

[settings]
slack_channel=#daily-standup

Choose the GitHub mode:

mode = org — fetch activity from all organization repositories
mode = repo — fetch from a single specified repo

[settings]
mode = org  ; or "repo"
organization = my-org-name  ; required if mode is "org"
owner = username            ; required if mode is "repo"
repo = repo-name  

---

Add GitHub Secrets

You'll need these for the agent to fetch data and send messages.

Go to:
Repository → Settings → Secrets and variables → Actions → New repository secret

Add the following:

Secret Name	Description
G_TOKEN	GitHub Personal Access Token
SLACK_BOT_TOKEN	Slack Bot Token
NOTION_TOKEN	Your Notion integration token
GROQ_API_KEY	API key for LLM summarization
DATABASE_ID	Notion database ID (where your tasks live)

---

B. GitHub Actions: Auto-Run Daily

This project includes a GitHub Action that can run the agent automatically every day.

To enable the daily auto-run:

1. Open .github/workflows/standup.yaml.
2. Uncomment the schedule: section and keep workflow_dispatch: to allow both daily automatic runs and manual triggers.

on:
  schedule:
    - cron: '42 16 * * *'  # Runs every day at 4:42 PM UTC
  workflow_dispatch:       # Allows you to run it manually

Customizing the Trigger Time

You can adjust the cron expression to change the time of day when the agent runs. For example, to make it run at 8 AM UTC, change it to:

You can change the schedule timing using this guide.

Microsoft Teams Integration

Require Microsoft 365 (business or enterprise) account to use the Teams integration for Webhook.

---

Run Locally

If you want to test before pushing:

python -m agent.standup_agent

Make sure the required environment variables are available. You can use a .env file or export them in your shell.

---

Example Slack Output

---

Extending the Agent

You can plug in other tools like Jira, Linear, or custom Git sources. Just update the fetchers inside:

tools/
├── github_fetcher.py
├── notion_fetcher.py
├── slack_poster.py

And adjust the prompt logic in:

agent/standup_agent.py

---

License

MIT License