AI-Code-Review Is Now Public
Putting myself out there
If you write software in 2025 and you’re not integrating AI into your workflow, you’re already behind.
Over the past year, I’ve been using a custom AI code review agent to streamline my development process. It flags architectural inconsistencies, tightens up type usage, and catches sloppy patterns that creep into even the cleanest codebases. It’s been useful enough that I kept improving it between projects—and now it’s out in the open.
👉 GitHub: bobmatnyc/ai-code-review
📦 npm i @bobmat/ai-code-review
Why Now?
1. Trust is earned, not claimed.
If I’m going to talk about agentic coding, I need to show what it looks like in practice. Not the polished demo version—real prompts, real critiques, real code. Transparency matters more than marketing. This repo is the actual tool I use, with all the quirks, caveats, and useful defaults.
2. Agentic workflows are the future.
We’re moving past autocomplete. AI-first programming means describing your intent and letting software build and review code with you. This tool is an example of that. It’s not just linting with extra steps—it thinks like an engineer. It considers design trade-offs, suggests restructuring, and occasionally calls out nonsense. That’s the bar for AI assistants in code.
3. It’s useful. Genuinely.
I use this tool on nearly every TypeScript project. It helps with:
Structural audits of component trees
Spot-checking for stale abstractions
Validating prompt-generated code
Keeping PRs focused and consistent
It’s not perfect. But it’s fast, opinionated, and helpful in the way a good senior engineer is: clear, practical, and never vague.
4. I want your feedback—and your pull requests.
This project isn’t frozen. If you find a prompt that fails or a pattern it should recognize, open an issue or submit a patch. If you’re working on AI + dev tooling and want to collaborate, reach out.
Bottom line:
AI code review should be more than “looks good to me” with emojis. This is a step toward making automated review smarter, more context-aware, and genuinely useful for developers who care about quality.
Let me know what you build with it.
Keeping code clean can be a real challenge sometimes. You might want to give mailsAI a shot for automating some of those tedious checks. It’s helped me catch issues early and made reviews less stressful overall.