Acuvis

Code reviewAvailable

Big pull requests, reviewed like small ones.

Acuvis is a code review IDE for GitHub. It reads every pull request before you do and gives it back as a walkthrough in review order, a map of what the change touches, and findings that come with the fix.

In detail

Screens from the live app

Start with the story, not the file list.

A summary of the whole change, then each part in the order to review it: what it does, the file to start in, and the findings that belong to it.

Acuvis walkthrough of a pull request, split into numbered parts with a summary and findings for each.

See what it touches before you read a line.

Parts laid out by area of the codebase, with what calls, uses and imports what. Select one to see its impact and the findings inside it.

Acuvis map view: clusters of changed files grouped by app and package, connected by call and import edges, with a findings panel.

Findings that come with the fix.

Security, privacy, correctness, performance, architecture and tests. Each finding says what is wrong, what it breaks and how to fix it, with a confidence score. Blocking ones are marked.

An Acuvis finding card describing a missing brute-force protection on an unlock endpoint, with impact and a suggested fix.

Then read the lines that matter.

Files grouped by part, in review order. Changed words are highlighted, viewed marks stick, and you can comment or suggest a change on any range of lines.

Acuvis files view with a syntax-highlighted diff and an inline comment containing a suggested change.

One clean review back on GitHub.

Pick a verdict and Acuvis posts your comments and the blocking findings as a single GitHub review. A summary comment and a merge check keep the pull request current.

The Acuvis submit dialog with a Request changes verdict and the comments that will be posted to GitHub.

How your data is handled

Application data is hosted in the EU, on OVH in France.
Each pull request is analysed in its own Firecracker microVM, destroyed when the analysis ends.
Only the diff needed to show your review is stored, scoped to your organisation.
Model inference runs through Fireworks AI with zero data retention and no training on customer prompts.
Every sub-processor is listed publicly, with 30 days' notice before a new one is added.

Security ↗Privacy policy ↗Sub-processors ↗

Availability

Available

Acuvis is live with self-serve signup. Install the GitHub App and the first five pull requests are free, no card needed.

Team plans start at $19 per active developer a month, billed annually. Reviewers, viewers and bots are free, and approved open-source repositories get 200 pull requests a month at no cost.

More from Lisovate

About the company →