Shipping My First MVP

Yesterday I wrote about starting in public. Today I shipped something real.

PR Risk Scorer — a GitHub App that analyzes every pull request and posts a risk score before your team merges.

The problem

Every team has a version of this story: a large PR lands on Friday afternoon, gets a quick review, merges, and something breaks over the weekend.

The signals were always there — auth files changed, no tests, a huge diff — but nobody had time to connect the dots before clicking merge.

I wanted a tool that does that automatically, right inside the PR conversation.

Step 1: Landing page

Before writing backend code, I built the landing page. Not because I needed marketing first — because it forced me to explain the product in one sentence.

The headline wrote itself: "Stop merging risky code on Friday afternoon."

The page shows exactly what the bot will post on a PR: a score, a breakdown of risk factors, and a recommendation. If I could not explain the output clearly on a landing page, the product was not ready to build.

Stack: Next.js, Tailwind, deployed to Vercel. Took one evening.

Step 2: The GitHub bot

The MVP is a GitHub App that listens to pull request events and leaves a comment with a risk score from 0–100.

It checks things like:

• Sensitive files touched — auth, payments, migrations (+30)
• No tests added when logic changed (+20)
• Large diff — more lines changed, more surface area (+15)
• Friday afternoon merge — because timing matters (+15)

Each factor adds to the score. The bot maps the total to a level — low, medium, or high — and gives a plain recommendation: request a senior review, wait until Monday, or merge with confidence.

No dashboard. No config UI. Just a comment on every PR, where the team already works.

What I deliberately left out

This is an MVP, not a v1:

• No custom rules or per-repo configuration
• No Slack or email alerts
• No historical analytics
• No paid plans yet — beta is free

The goal was one loop: install the app → open a PR → see a useful comment. That loop works.

Numbers

| Metric | Value | | --- | --- | | Time to MVP | ~2 days | | MRR | $0 | | GitHub installs | early beta | | Lines of backend | ~400 |

What's next

• Waitlist → beta invites
• More risk signals (dependency changes, hot paths, reviewer load)
• Repo-level baselines so scores are relative, not absolute

If you manage a team that merges on Fridays, join the waitlist or install the beta on GitHub.

Building from Tashkent, shipping to the world — one PR at a time.