I built an AI tool that enforces your team's coding standards on every PR automatically

Dev.to / 4/11/2026

📰 NewsDeveloper Stack & InfrastructureTools & Practical Usage

共有:

Key Points

The article describes Solon AI, a GitHub App that automatically enforces a team’s custom coding and review playbook rules on every pull request to prevent missed checks and reduce tech debt.
Solon AI works by connecting to GitHub repositories, letting teams configure rules via a dashboard UI (without config files), and then reviewing PRs automatically with comments and saved review history.
The tool supports configurable severity levels (blocking vs warning) so teams can decide how strongly standards violations should affect merges.
It includes an MCP server that allows standards checks to run directly from tools like Claude Code or Cursor before a PR is opened.
Solon AI lists an ethics-focused approach (e.g., diff-only processing, avoiding user_id in reviews, and preserving human override), with pricing that offers a free tier and a $29/month Pro plan.

Every team has a checklist of things they should verify in every PR — accessibility compliance, security patterns, code quality rules — but manual checks get forgotten. Tech debt accumulates.

I built Solon AI to solve this. It connects to your GitHub repos as a GitHub App and automatically enforces your team's playbook rules on every pull request.

How it works

Connect your repo to Solon AI
Configure your playbook rules in the dashboard (no config files needed)
Open a PR — Solon reviews it automatically and posts a comment

Example output

I added this intentionally bad code to a test PR:

<img src="photo.jpg">
<button>Click me</button>

Solon caught both violations immediately:

What's built

GitHub App webhook — zero setup, works automatically on every PR
Visual playbook configuration — no config files, configure rules in the dashboard UI
BLOCKING/WARNING severity — teams decide what blocks vs warns
Review history dashboard — every review saved, searchable, deletable
MCP server — check standards directly from Claude Code or Cursor before opening a PR
Ethics by design — no user_id in reviews, diff-only processing, human override always maintained

Pricing

Free tier: 25 reviews/month, 1 repo
Pro: $29/month flat — unlimited reviews, unlimited repos, no per-seat fees

Stack

Next.js 15, TypeScript, Neon Postgres, Claude Haiku, GitHub App webhooks, Stripe, Vercel

Looking for

Engineering managers at teams of 10–50 developers tired of catching the same PR issues every sprint. Happy to set you up on Pro free in exchange for honest feedback.

👉 solonreview.dev