It's Monday and I Can't Remember What I Changed on Friday
How gitaiflow turns an unfinished working tree into a plain-English note you can read in ten seconds — and keep for next time.
CB
You know the feeling. It's Friday afternoon, something urgent pulls you away, and you leave a half-finished change sitting in your working tree. On Monday you open the repo, run git status, and get fourteen modified files and three untracked ones.
File names tell you where you were. They don't tell you what you were doing. git diff does, but at six hundred lines, reading it is a task of its own. So you reach for the usual fixes — a stash named wip, a TODO comment, a vague memory — and none of them really work.
What I wanted was a short note: what changed, in which areas, in plain language, without me having to write it. That's the first thing I use gitaiflow for.
1. Summarize the Unfinished Work, Not Just Commits
gitaiflow compares your working tree against HEAD. That covers staged changes, unstaged changes to tracked files, untracked files, and deleted files — everything that would go into a commit if you made one right now.
gitaiflow --path . --change-summary
The output is a commit-style title plus a few bullets. Something like this (the wording depends on the model you use):
billing: add retry handling for failed invoice sync - Added retry with backoff in sync.py when the payment API times out - invoices/views.py now returns a pending status while a retry is in flight - Removed the unused legacy_export.py helper
Deleted files are part of the summary, which matters more than it sounds. Removed files are the easiest thing to forget you did. To focus on one area, point --path at a folder, like --path billing/.
2. Leave Yourself a Note Before You Leave
Every run also saves a timestamped JSON file in a change-summary/ folder at the root of the repo, grouped by date. That folder is what makes Monday easy:
gitaiflow --last-summary
--last-summary prints the most recent saved summary straight from disk. No git diff, no AI call, no tokens spent. If nothing has been saved yet, it tells you so instead of generating one.
One honest caveat: --last-summary shows the tree as it was the last time you ran it. If you kept working after that, run a fresh summary. It won't cost you much — if a group of changes is identical to one already summarized, gitaiflow reuses the earlier summary instead of sending it to the AI again.
3. Keep the Noise Out
Generated files and build output can drown the real change. --skip takes folders, exact paths, and file types:
gitaiflow --path . --change-summary --skip dist/ "*.svg" "*.json"
Want the Full Picture?
This post covers the Friday-to-Monday habit. For every option, provider setup and configuration detail, see the gitaiflow documentation ↗.
What I'd Tell You If Your Mondays Look Like Mine
- A note written when you stop beats a better note written after you've forgotten. The value is in capturing the context while it's still in your head.
- Make the note cost nothing. If leaving yourself a summary takes more than one command, you'll skip it on exactly the Fridays you need it.
- Treat a generated summary as a reminder, not a record. Skim it before you rely on it — it's built from the diff by a model, and models can get details wrong.