Standards first.
Facts checked.

aicodereview.io answers one question: what should an AI code reviewer actually be able to do in 2026? The answer is 9 standards — an engineering baseline for the category — and everything else here applies them: a directory of 27 tools scored against those standards, 30 blog articles, 10 guides, and a 57-term glossary.

What makes this different from other tool directories

No star ratings pulled out of the air, no affiliate links, and no paid placements. Every tool is mapped against the same nine standards with four values — documented, partial, not offered, undocumented — each carrying a note drawn from the vendor's own documentation and a date it was last checked. The coverage score is arithmetic over that table, published in full so you can recompute it or reject it.

The score measures what a vendor documents, not what we measured in a lab. We do not claim to have benchmarked how many real bugs each tool finds on your codebase, because nobody can do that honestly from the outside.

How facts are verified

Pricing, licensing, self-hosting and platform support come from the vendor's own pricing page, documentation or public repository, and each tool carries the date it was last checked — currently 2026-08-11 across the directory. Where a vendor publishes no number, the field says so rather than carrying an estimate. Vendor-published benchmark figures are attributed and labelled as such, and never folded into a score.

The complete dataset lives in the repository as one JSON file per tool. A wrong or stale claim is a bug — open an issue and it gets fixed.

Who runs it

This site is operated by the team at Kodus, who build one of the tools in the directory. Nothing here is for sale: no paid placements, no affiliate links, and no arrangement under which a ranking or a criticism can change hands.

Kodus is scored with the same rubric as everything else — it lands on 6.5/9, with the 4 standards it does not fully meet listed openly on its profile.

A directory run by a vendor deserves scepticism, and the answer to that is not a disclaimer — it is making every input checkable. The whole site, dataset included, is open source on GitHub, so any thumb on the scale is visible in the commit history. The scoring policy is on the methodology page.

For AI engines and agents

This site welcomes AI crawlers — being read and cited is part of how it reaches people. Machine-readable versions live at /llms.txt and /llms-full.txt, and raw markdown mirrors exist for every blog post (/blog/<slug>.md), every tool profile (/tools/<slug>.md) and every glossary term (/glossary/<term>.md).