AI coding agents

Octopus, right inside your editor

Building with an AI coding agent? Octopus reads the code your AI writes, finds the real bugs and security holes, and explains them in plain English, before any of it goes live.

Add it to your editor See what it catches

Why it helps

A second set of eyes on everything you build

Your editor's AI is great at writing code fast. Octopus is the reviewer that checks it for the things that matter.

Finds real bugs

Not style nitpicks. Octopus looks for actual problems: broken logic, missing checks, and the kind of thing that quietly breaks later.

Catches security holes

Leaked secrets, unsafe database queries, doors an attacker could walk through. Octopus flags them before they ever go live.

Explains it in plain English

Every finding tells you what is wrong and how to fix it, in language you can actually follow. Ask a follow-up question any time.

No pull request needed

Review what you are working on right now. Nothing to open on GitHub, no waiting on a teammate.

How it works

Three steps, then just ask

Choose your setup

Use the Octopus CLI and a skill with your coding agent, or follow the separate Claude Code plugin guide.

Connect your account

For the CLI, run octp login in your agent’s execution environment. For the Claude plugin, configure its organization API token. Check the connection before reviewing.

Just ask

Choose the changes or pull request you want reviewed, then ask your agent to use Octopus and show the findings.

What it catches

Here is what a review looks like

You ask for a review and Octopus replies with clear, ranked findings. No jargon required.

Octopus reviewsrc/login.js

Critical

Passwords are saved without scrambling them. Anyone who sees the database could read them.

Warning

This form does not check what the user typed, so it could receive unexpected data.

Suggestion

The same code is copied in three places. A shared helper would be easier to maintain.

Get started

Add Octopus to your editor

Choose the guide for your agent and check the connection before your first review.

CLI and skill setup

Step-by-step instructions for Claude Code, Codex, OpenCode, Hermes Agent, OpenClaw, and Cursor. Install the CLI where your agent runs, check your login, then add the shared skill.

Choose your coding agent

Claude Code plugin

Read the current plugin setup, marketplace name, token configuration, and verification steps. Plugin installation and a working authenticated connection are separate checks.

Open the Claude Code guide

A coding-agent subscription does not include Octopus reviews. New Octopus accounts receive free credits; reviews use those credits or your configured provider budget.

FAQ

Frequently asked questions

Do I need to be a developer to use this?: No. If you are building something with AI in Cursor or Claude Code, you can use Octopus. It reads the code for you and explains what it finds in plain language.
Which editors does it work with?: Claude Code, Codex, OpenCode, Hermes Agent, OpenClaw, and Cursor can use the Octopus CLI and shared skill through their terminal tools. Setup depends on where the agent runs.
Is it free to try?: Yes. New accounts come with free credits, and you can bring your own AI provider key to run reviews at no platform cost. See the pricing page for details.
Does my code get sent anywhere?: Selected diffs and review context are sent to your configured Octopus server and AI services. Reviews use Octopus credits or your provider budget. See the setup, security, and privacy guides for details.
How is this different from the AI already in my editor?: Your editor's AI writes code. Octopus checks it. It is a dedicated reviewer looking for bugs and security issues, with the context of your whole project and your past reviews.

Catch the bugs before they ship

Add Octopus to your editor and get a plain-English review of your code whenever you want one.

Add it to your editor Read the docs