Last updated 19 November 2026

Privacy

Git Unpacked turns a GitHub organisation's year into a story. This page says exactly what it reads, what it keeps, who can see it and how to delete it.

The short version

  • Read-only access to the repositories your organisation chooses. Nothing is ever changed on GitHub.
  • Your code, commit messages, and the titles and text of pull requests and issues are never stored or used.
  • No ads, no analytics, no tracking.
  • No names are sent to the AI that writes captions: it only sees anonymised facts.
  • Disconnecting an organisation deletes everything held about it.
  • Your own year is shown only to you unless you share it.

What we read from GitHub

An organisation owner installs the Git Unpacked GitHub App and picks which repositories it can see. It asks for read access only:

Repository metadata
Repository names, whether each is private, archived or a fork, and its languages.
Repository contents
Commit history only: each commit's id, author name and email, time, and whether it's a merge. GitHub's commit list includes messages; we don't store or use them. We never read files.
Pull requests
Number, author, labels, draft status and when each was opened, merged or closed. For reviews: who reviewed, whether they approved or asked for changes, and when.
Issues
Number, author, labels and when each was opened or closed.
Organisation members
Who is in the organisation and its teams, and who owns it. This decides who can see the organisation here.
Email addresses (your account)
Your email, to sign you in.

Owners can also leave repositories out of the report; activity in those isn't collected.

When you open your own year, we also read your GitHub contribution summary using your sign-in: your daily contribution counts, your totals of commits, pull requests, reviews and issues, and the public repositories you committed to, with their languages. Work in private repositories arrives only as a count: we never learn which private repositories it was in.

What we keep

  • The organisation: its name, avatar and settings, its repositories, its teams (name, members and repositories) and its member list (GitHub user id, username and whether each person is an owner).
  • The year's activity in included repositories, as listed above, and the people behind it: name, username, GitHub user id, whether they are a bot, and the emails on their commits, which we use to recognise the same person across commits.
  • Reports: the numbers, award winners and captions computed from that activity, and a personal recap for each contributor with a GitHub account.
  • You, if you sign in: your GitHub name, username, user id, email and avatar, the token GitHub issues when you sign in, and for each session your IP address and browser, to keep you signed in securely.
  • Your own year: the summary of your contribution calendar described above, refreshed when you open it.
  • Your choices: your answers to guess questions in team stories, and whether you have hidden yourself from awards and top lists.
  • In your browser only: the sign-in cookie, and your theme, sound and autoplay preferences.

Who can see what

  • Owners (the organisation's owners on GitHub) see the dashboard, drafts, stats and settings.
  • Members of the organisation on GitHub see the published story and stats. Stories name colleagues and included repositories.
  • Your personal recap is shown only to you: not to owners, not to teammates.
  • Your own year is shown only to you. If you share it, anyone with the link can play it: it shows your username, your numbers and the public repositories you committed to. Stop sharing and the link stops working.
  • Your guesses are never shown to anyone. Team results show counts only, and only once at least 5 people have answered.
  • Public share links, created when an owner publishes, show organisation-wide numbers only: no people and no repository names. They ask search engines not to index them.
  • Opting out: anyone in the organisation can hide themselves from awards, the top 5, the dream team and guess options. Their work still counts in the totals.

AI captions

The jokes under each card are written by an AI model run by Groq. For each card it receives one line of facts, such as a number, a month, a language or an award name, with every person, repository, team and organisation name replaced by a placeholder like {person_1}. Real names are put back on our side. It never receives code, commit messages, or pull request and issue text. Every caption is checked before use, and any that fails keeps a hand-written line.

Companies that handle data

GitHub
The source of the activity data, and sign-in.
Vercel
Hosts the app.
Neon
The database where everything below is stored, in the United States (AWS us-east-1).
Inngest
Runs the background jobs that collect activity. It sees organisation and repository names, internal ids and counts, not the activity itself.
Groq
Writes story captions from anonymised facts (see below).

We don't sell data or share it with anyone else.

How long we keep it, and deleting it

  • Organisation data is kept while the organisation stays connected. Each unpack refreshes the year, and commits removed from GitHub are removed here too.
  • An owner can disconnect the organisation from the dashboard, or uninstall the GitHub App on GitHub. Either deletes everything held about the organisation: activity, reports, recaps, answers and settings.
  • You can stop sharing your own year at any time from the dashboard.
  • To delete your own sign-in account, including your own year, or to get a copy of what we hold about you, email us.

Contact

Questions or requests about your data: anibiseun@gmail.com. When what we collect changes, this page changes with it, and the date at the top says when.