Mesrai

AI code review that reads your repo, not just the diff

5.0•3 reviews•

5 followers

AI code review that reads your repo, not just the diff

5.0•3 reviews•

5 followers

Visit website

Code Review Tools

Most AI code reviewers look at the diff. Mesrai reads the whole repo. Before reviewing each PR, Mesrai builds an AST graph of your codebase imports, type relationships, call chains across files. Five specialist agents run in parallel: security, performance, architecture, bugs, and your team's plain-English rules. Results: 86% critical defect catch vs 57% for CodeRabbit on the same LLM. BYOK — bring your own Anthropic/OpenAI/Vertex key. No markup. $6/dev/mo. 14-day free trial. No credit card.

Free Options

Launch tags:Developer Tools•Artificial Intelligence•GitHub

Launch Team / Built With

Render Sandboxes: a computer for every agentGive every agent its own machine. Starts in under a second

Promoted

Maker

📌

Hey Product Hunt 👋 I'm Kajal, a full-stack developer from India. I built Mesrai because I kept watching the same thing happen on every team that adopted Copilot or Cursor: PR volume went up, review bandwidth didn't. Senior engineers started drowning in queues. Junior devs waited days for feedback. Things slipped through. The core problem with most AI reviewers: they only see the diff. That misses an entire class of bugs — the ones that require understanding how the changed code connects to the rest of the system. Mesrai builds an AST graph of your repo before each review. Five agents run in parallel: security, performance, architecture, bug detection, and your own custom rules (written in plain English, not regex). We ran a public benchmark — 24 PRs across Supabase, Apache Airflow, and HashiCorp Vault. Both Mesrai and CodeRabbit ran on the same LLM (claude-opus-4-7) with default settings: → Overall catch rate: Mesrai 75%, CodeRabbit 63% → Critical findings: Mesrai 86%, CodeRabbit 57% We published every miss too, not just the catches: mesrai.com/compare/coderabbit Honest gaps I want to flag upfront: → Per-folder rule scoping is on roadmap — not live yet → No permanent free OSS tier (CodeRabbit has this, we don't) → Self-hosted is Enterprise tier 14-day free trial, full features, no card: mesrai.com Happy to answer anything — about the AST approach, the benchmark methodology, BYOK setup, or why we built this from India. 🙏

Report

4mo ago

Forum Threads

p/mesrai

•

3mo ago

Solo dev building Mesrai (AI code review) — let's support each other!

Hey everyone!

I'm the solo developer building Mesrai. I don't have a giant team, VC funding, or a marketing budget. I'm building this entirely on my own because I wanted an AI code reviewer that catches deep architectural bugs and security flaws automatically on pull requests.

View all

5.0

Based on 3 reviews

Review Mesrai?

Pros

byok pricing (2)

free trial, no card (2)

multi-agent specialist reviewers (2)

ticket validation (2)

whole-codebase dependency graph (2)

Cons

could add more integrations with other tools (2)

social proof (2)

Pros

Cons

Reviews

Most Informative

•2 reviews

What's great

whole-codebase dependency graph (2)multi-agent specialist reviewers (2)byok pricing (2)ticket validation (2)free trial, no card (2)

Mesrai fits into the way developers already work, not only the PR page. The CLI lets me run a review before I push, and it can be wired into a pre-push hook, so many issues are caught before a teammate ever sees the PR. It also works with the AI coding agents I use, and there's a VS Code extension.

What I liked most:

Pre-push local review. Review uncommitted changes from the terminal and catch things early.
Custom rules in plain English or YAML (Mesrai Playbook), applied per org, repo or directory, so reviews follow our team's standards.
Severity-ranked inline feedback with the reasoning and a suggested fix, which makes it quick to decide what matters.
Works on GitHub, GitLab, Bitbucket and Azure DevOps, so it isn't tied to one host.
Fast onboarding. Connect the repo, add a key, push a PR. It took a couple of minutes.
Privacy. Code is processed in ephemeral sandboxes, isn't stored, and isn't used for training.

What needs improvement

could add more integrations with other tools (2)social proof (2)

It's a newer product, so a public changelog and a visible roadmap would help teams see where it's heading. Review analytics (trends in issues caught per repo or per team) would also be useful, as would more detailed examples in the docs for writing Playbook rules.

vs Alternatives

Qodo (formerly CodiumAI)

CodeRabbit

Graphite

cubic

I compared Qodo, Graphite and cubic. They are good at in-PR comments, but I wanted a tool that also works locally before the PR exists, with rules my team can write ourselves and control over which model and key it runs on. Mesrai covered all three.

Ratings

Ease of use

Reliability

Value for money

Customization

Report

10 views1mo ago

•1 review