Lakshmi Sravya Vedantham

Posted on Feb 23

I Built a CLI That Auto-Generates My Daily Standup Notes from Git History

#productivity #python #developer #opensource

The Problem: Nobody Remembers What They Did Yesterday

Every morning, the same ritual. Standup starts in 5 minutes. You stare at Slack trying to reconstruct what you worked on. You check your browser history. You scroll through git log. You write something vague like "continued work on the feature."

I got tired of this. My git history already has everything — what I changed, when, and why. Why am I manually summarizing it every day?

So I built dev-journal.

3 Lines to Get Started

pip install dev-journal
dev-journal collect --since 2026-02-17
dev-journal standup

That's it. Your standup is ready to paste.

What It Generates

Daily Standup

## Daily Standup — February 23, 2026

### What I did yesterday:
- **[my-app]** feat: implement rate limiting middleware (4 files, +112/-8)
- **[my-app]** fix: correct null check in auth handler (2 files, +15/-6)
- **[api-client]** chore: bump httpx to 0.27 (1 file, +3/-3)

### What I'm working on today:
- Continue work on branch: feature/oauth2-integration

### Blockers:
- None

---
*Stats: 3 commits | 7 files | +130/-17 lines | repos: my-app, api-client*

Weekly Sprint Summary

Aggregates an entire ISO week: total commits, lines changed, per-repo breakdown, daily heatmap, and top highlights by impact.

Blog Post Draft

Generates Dev.to-compatible markdown with YAML frontmatter, narrative sections per repo, stats, and placeholder sections for "Lessons Learned." (Yes, I used dev-journal to help write this post.)

How It Works

dev-journal has three collectors and three generators:

Collectors:               Generators:
┌──────────────┐         ┌──────────────────┐
│ git log      │───┐     │ standup (daily)   │
│ file mtime   │───┼────>│ weekly (sprint)   │
│ shell history│───┘     │ blog (Dev.to)     │
└──────────────┘         └──────────────────┘
        │                         │
        ▼                         ▼
   SQLite DB              text / markdown / json

Git Collector

Parses git log --stat output and classifies each commit's intent:

Prefix	Intent
`feat:`	FEATURE
`fix:`	FIX
`refactor:`	REFACTOR
`docs:`	DOCS
`test:`	TEST
Everything else	CHORE

It also extracts file counts, insertions/deletions, and normalizes timezone-aware timestamps.

File Collector

Scans for recently modified files grouped by language and directory. Respects .gitignore patterns and skips common junk (__pycache__, node_modules, etc.).

Shell Collector (Opt-in)

Parses zsh/bash history for developer commands (git, pytest, docker, make). Disabled by default. Sensitive commands containing password, token, secret, or api_key are automatically filtered and never stored.

Architecture

src/dev_journal/
├── cli.py               # Click CLI: 7 commands
├── config.py            # TOML config at ~/.dev-journal/
├── storage.py           # SQLite with full CRUD
├── formatter.py         # Rich terminal output
├── collectors/
│   ├── git_collector.py     # git log parser
│   ├── file_collector.py    # mtime scanner
│   └── shell_collector.py   # history parser
└── generators/
    ├── standup.py           # daily notes
    ├── weekly.py            # sprint summary
    └── blog.py              # Dev.to draft

Privacy First

This was a core design principle:

All data is local. SQLite at ~/.dev-journal/journal.db. Nothing leaves your machine.
Shell history is opt-in. Disabled by default. You explicitly enable it.
Sensitive commands are filtered. Anything matching password, secret, token, api_key is skipped.
You control what's tracked. Configure ignored extensions, directories, and patterns.

The Interesting Technical Bits

Commit Intent Classification

I wanted the standup to group commits by what kind of work they represent, not just list them chronologically. The classifier uses conventional commit prefixes first, then falls back to keyword analysis in the commit message:

Messages containing "add", "implement", "create" -> FEATURE
Messages containing "fix", "resolve", "patch" -> FIX
Messages containing "refactor", "restructure", "clean" -> REFACTOR

Standup "Today" Inference

The hardest generator problem: predicting what you'll work on today. dev-journal looks at:

Open branches (not merged to main)
Recent WIP commits
Commits from today that look incomplete (small changes, no tests)

It's not perfect, but it beats staring at a blank text box.

Numbers

67 tests passing
74% coverage
Python 3.9-3.12 matrix CI
3 output formats: text, markdown, JSON

Try It

pip install dev-journal
dev-journal collect --since $(date -v-7d +%Y-%m-%d)
dev-journal standup --copy  # copies to clipboard

Star it on GitHub: github.com/LakshmiSravyaVedantham/dev-journal

How do you prepare for standup? Let me know in the comments if you've tried automating it.

DEV Community