About Critiq
We believe review should feel like a careful teammate.
Critiq is the trust layer for AI-assisted development. Transparent rules, inspectable evidence, and feedback you can reproduce before it ever blocks a merge.
“Developers are shipping with AI more than ever. They deserve review they can actually trust.”
Why we started
Why we started
AI made writing code faster than ever. It did not make reviewing it any clearer. Critiq exists to close that gap.
01. The problem
AI-generated code is accelerating development, but most review tools feel like black boxes: noisy feedback, unclear reasoning, and no way to verify what actually triggered a finding.
02. Our answer
Critiq pairs deterministic, rule-based analysis with optional AI insight. Every finding names the rule that fired, the evidence behind it, and the part of your system it affects.
03. Built in the open
The core engine, the public CLI, and the OSS rules catalog all live in public repositories. Run scans locally, read the rules, fork them, and contribute fixes without waiting on a vendor roadmap.
What we value
How we show up
Six traits shape every rule we write, every finding we surface, and every screen we ship.
- Smart
Deep analysis
Findings come from explicit rules and real evidence, not vague summaries you cannot verify.
- Calm
Signal over noise
We prefer fewer, sharper findings over dashboards full of low-value alerts nobody reads.
- Honest
Useful feedback
Every result explains what triggered it and what to do next, in plain language.
- Human
Developer-first
Critiq fits the local workflows, CI, and review habits teams already have.
- Crafted
Quality over hype
We ship inspectable tools and documentation that match what runs today, not launch-day promises.
- Warm
An inviting experience
The site and the CLI should feel approachable, even when the findings are serious.
How we operate
How we work
Critiq is open source first. Premium rule packs and the hosted product are separate layers built on the same contracts, never a replacement for the open toolchain.
We optimize for trust. A finding you can read, test, and reproduce is worth more than a confident guess you have no way to check.
- Deterministic rules you can read, validate, and test with fixtures.
- Local-first scanning that keeps your checkout on your machine or CI runner.
- Documentation and examples that match what ships in the catalog today.
Open by default
Built in the open, not behind a sales call.
The engine, CLI, and rule catalog are public. Read the rules behind every finding, run them locally, and tune them to the way your team actually reviews code.
- 435+
- Public rules in @critiq/rules
- 8
- Languages analyzed deterministically
- 100%
- Local by default: your code stays with you
Explore the open toolchain
Install the CLI, browse the rules catalog, and decide from real output, not marketing slides.