AI coding security · Find your starting point

A useful first check.
For the way you use AI.

Review what your agents can reach, how they are configured, and the code they change. Choose the problem you want to work on.

Explore the reviews

Available workflows

What would you like to check?

Choose a starting point.
Each check has a clear scope.

Editor extension

Understand what your agents can access

Review the tools, hooks and repository access configured in your environment. Find settings that deserve a closer look before your next session.

Configuration findings, not a guarantee that actions are blocked.

Set up a local review
Editor extension

Check your MCP server setup

See which MCP servers are configured and review unexpected connections or exposed credentials in server arguments.

Configuration review, not live traffic monitoring.

Review my MCP setup
Browser tool · no account needed

Check AI-written code before you ship

Try a browser review of a code snippet and inspect potential security issues. Use the findings to guide your next check.

A useful starting point, not proof that code is safe.

Try code review
Editor extension

Catch risky agent edits as they land

See findings on supported agent edits and review a change you may want to undo, inside your editor.

Checks after a write; does not block the write.

Review edits in my editor
Editor extension

Spot suspicious dependency names

Check for suspicious package names in your project dependencies as part of your local environment review.

Name checks do not establish that a package is trustworthy.

Check my dependencies
Browser questionnaire

Take stock of your AI security practices

Work through a browser-based questionnaire about your AI setup and see which areas deserve attention.

Based on your answers; it does not inspect your environment.

Start a self-assessment

Get the context

Know what a finding can tell you.

Use these guides to understand the checks before relying on a result.

Before you start

A few practical answers.

Can I review code without installing an extension?

Yes. The browser review checks a code snippet without an account. Use the editor extension to review the configuration of your local agent setup.

Does an environment review stop every risky action?

No. A configuration review identifies findings. Checks on supported agent edits happen after a write. The coverage guide explains what each check does and where its limits begin.

Which AI tools can I use DeepSweep with?

Start with the editor chooser to find the current setup instructions for your editor. The coverage guide lists the available checks. Support for an editor does not mean every model, tool or action is covered.

Is this a directory of every AI product?

This is a guide to DeepSweep's available AI coding review workflows and learning resources. It helps you choose a useful first check for the problem in front of you.