Skip to main content
You’ve been shipping code for weeks (months? years?), but when someone asks “what have you been working on lately?”—you draw a blank. You know you built something important, but the details are fuzzy. That’s where repr comes in. It reads your git history and turns your commits into professional narratives. Here’s how to get your first stories generated in about 5 minutes.

​
What You’ll Need

  • The repr CLI installed (see Installation if you haven’t yet)
  • A folder with some git repositories (like ~/code or ~/projects)
  • (Optional) Ollama running locally if you want 100% offline generation
Let’s go.
1

Scan your repos

Point repr at your code directory and let it discover what you’ve been working on.
repr init ~/code
Repr will scan for git repositories and show you what it found:
Found 12 repositories
✓ myproject (143 commits) [Python]
✓ frontend-app (87 commits) [TypeScript]
✓ api-gateway (56 commits) [Go]
✓ docs-site (22 commits) [Markdown]
...

Track these repositories? [Y/n]
Hit enter to confirm. Your repos are now tracked, but nothing has left your machine yet—this is all local.
2

Take a quick look around

Before generating anything, let’s see what you’ve been up to this week.
repr week
You’ll get a quick summary like:
📊 Last 7 days across 4 repositories

myproject (12 commits):
  • Add OAuth2 integration
  • Implement Redis caching
  • Fix race condition in auth flow

frontend-app (8 commits):
  • Redesign settings page
  • Add dark mode toggle

Total: 20 commits, 8 files changed
This is just a preview—nothing saved yet. But it gives you a sense of what repr will work with.
3

Generate your first stories

Now for the magic. Let’s turn those commits into professional narratives.
repr generate --local
If you don’t have a local LLM installed, repr will guide you through setting up Ollama or adding an API key (OpenAI, Anthropic, etc). Your choice.
You’ll see repr processing your work:
Generating stories (local LLM: llama3.2)...

myproject
  • Built OAuth2 integration with Google and GitHub providers
  • Implemented Redis caching layer for session management
  • Fixed race condition in user authentication flow

frontend-app
  • Redesigned settings page with improved UX
  • Added dark mode toggle with system preference detection

Generated 5 stories
Stories saved to: ~/.repr/stories
That’s it. You’ve got professional summaries of your work, stored locally as JSON files.
4

Check out your stories

See what got generated.
repr stories
You’ll get a list with IDs, dates, and titles:
01JAXK... Built OAuth2 integration with Google/GitHub    Jan 5, 2026
01JAXL... Implemented Redis caching for sessions          Jan 5, 2026
01JAXM... Fixed race condition in auth flow               Jan 4, 2026
01JAXN... Redesigned settings page with improved UX       Jan 3, 2026
01JAXO... Added dark mode with system detection           Jan 2, 2026
Want to see the full story? Pick an ID and view it:
repr story view 01JAXK...
You’ll see a markdown-formatted story with context, technologies used, and impact.

​
What Just Happened?

You just turned your messy git history into clean, professional narratives. Here’s what repr did:
  1. Scanned your commits - Read commit messages, diffs, and file changes
  2. Grouped related work - Identified logical units (features, fixes, refactors)
  3. Generated context - Used an LLM to write professional summaries
  4. Saved locally - Everything lives in ~/.repr/stories on your machine
Nothing was uploaded anywhere. Your code stays on your machine. The only network call was to your local LLM (if you used --local) or to your API provider (if you used your own key).

​
What’s Next?

Now that you’ve got stories, here’s what you can do: The power move? Set up git hooks and let repr track everything automatically. Then, once a week, review and publish your best work. Five minutes a week to maintain a living portfolio of what you’ve actually built.