Octopus Code Review installs statistics

Octopus Code Review uses RAG (Retrieval-Augmented Generation) and vector search to deliver AI-powered code reviews with full codebase context. Unlike simple diff-based tools, Octopus indexes your entire codebase using Qdrant vector database, retrieves relevant context for every PR, and delivers severity-rated inline comments directly on your pull requests. Privacy-first: your code is processed in-memory and never stored.

How It Works

When a pull request is opened, Octopus fetches the diff, retrieves relevant context from your indexed codebase using vector embeddings, and sends the combined analysis to an LLM (Claude or OpenAI) for evaluation. The result: context-aware, severity-rated inline comments posted directly on your PR.

Key Features

• RAG-Powered Reviews: Full codebase context via Qdrant vector search, not just diff analysis • Severity-Rated Findings: Critical, Major, Minor, Suggestion, and Tip classifications • RAG Chat: Query your codebase using natural language with vector search and reranking • Knowledge Base: Feed your organization's standards and docs to customize reviews • CLI Tool: Run reviews locally via @octp/cli for terminal-based workflows • Analytics: Track review quality, token usage, and cost metrics • Self-Hostable: Deploy on your own infrastructure with Docker

Integrations

Works with GitHub and Bitbucket. Slack and Linear integrations available for notifications. Bring Your Own Key (BYOK) support for Claude and OpenAI.

Privacy & Security

Your source code is processed in-memory only and never permanently stored. Only vector embeddings are persisted for search. Self-hosting keeps everything on your infrastructure.

Open Source

Octopus is open source under the Modified MIT license. Free to self-host, with a cloud option available for teams who prefer a managed solution.

Description from the GitHub Marketplace listing.