rockyreviewsView on GitHub

AI assisted code review · Human controlled

Human reviews
enhanced by AI.

Rocky reads the pull request, sorts the risk, and drafts findings. Your team gets a clear place to start, and the final say on every review comment.

Explore the open source project See how Rocky works
Open source coreHost it yourselfBuilt for pull requests

Backed by mY self determination.

Every finding is a draft until a reviewer decides what to do with it.

A clearer review, start to finish

Less searching.
More signal.

A pull request is more than a stack of changed files. Rocky helps you understand what changed, where to look, and which questions deserve a human answer.

01

Know where to look

A file list ordered by risk gives reviewers a clear path through the pull request. Start with the changes most likely to matter.

02

Keep every call yours

Rocky drafts a finding. You accept it, edit it, or dismiss it. Nothing written by AI is posted without your approval.

03

See how changes connect

A change map makes the shape of a pull request legible, from the files it touches to the paths through the system.

04

Ask the pull request

Ask a question about the change and get an answer grounded in its diff, findings, and project context.

The change map

See the shape
of the change.

Follow the path from an entry point through the files a pull request touches. A visual map helps reviewers connect code that a flat file list can hide.

Follow the review flow

A review stays in your hands

Three steps.
One human decision.

Rocky prepares the context. Your reviewer chooses what belongs in the conversation.

  1. 01

    Connect a pull request

    Rocky follows the review on GitHub and builds an analysis for the current revision.

  2. 02

    Review the signal

    Scan the risk summary, move through files in order, and explore the change map.

  3. 03

    Make the call

    Accept, edit, or dismiss each draft. Your review stays yours from first pass to final comment.

Your team stays in control

AI can suggest.
Only you can post.

Rocky creates review drafts, never automatic comments. Accept a draft as written, edit it to fit your voice, or dismiss it with a reason. Every action stays attached to the finding.

01 Human approval before anything is posted

02 Review findings with their source and context

03 Project memory helps future reviews learn from decisions

Straight answers

Rocky Reviews
FAQ.

What developers should know about Rocky’s AI code review workflow.

What is Rocky Reviews?

Rocky Reviews is an open source AI code review tool for GitHub pull requests. It organizes changed files by risk, drafts review findings, and helps developers understand how code changes connect.

Does Rocky post review comments written by AI automatically?

No. Rocky prepares drafts for a person to review. A reviewer must accept or edit a finding before it can be posted; they can also dismiss it.

How does Rocky help review a pull request?

Rocky creates a file list ordered by risk, summarizes the pull request, drafts findings with file context, and visualizes relationships between changed code in a change map.

Can I host Rocky Reviews myself?

Yes. Rocky’s open source core is licensed under version 3 of the GNU Affero General Public License and can run on your own infrastructure. Visit the public GitHub repository for the code and setup guidance.

What does Rocky use as the source of truth for pull requests?

GitHub is the source of truth for pull requests, comments, and approvals. Rocky stores derived review data that can be rebuilt from GitHub.

Open source core · GNU Affero General Public License

Bring Rocky
to your workflow.

Explore the code, run Rocky on your own infrastructure, and help shape a more thoughtful way to review pull requests.

Browse the repository Public project · GitHub