🥋 Samurai Agent

Prevent JS/TS regression bugs by reviewing AI code edits inside your IDE — before they break your app.

TL;DR

Samurai Agent watches AI-generated code edits and immediately flags risky changes.
With one click, you get a focused, JS/TS-aware code review to catch regressions before you ship.

The Problem

AI coding tools like Cursor and Claude write code fast —
but they often introduce silent regressions, such as:

Race conditions
Risky async changes
Behavior-changing refactors
“Looks correct” fixes that break edge cases

These issues are hard to notice in review and expensive to debug later.

The Solution

Samurai Agent acts as an in-IDE guardrail for AI edits.

When AI changes your code:

Samurai Agent detects regression risk
A warning appears directly in your IDE
With one click, it runs a focused code review
You catch issues before they ship

No extra prompts. No copy-paste. No context switching.

Why Samurai Agent Works

Generic AI is language-agnostic.
Samurai Agent is JS/TS-specific.

Uses real JS/TS regression bug data
Finds last-mile bugs generic AI misses

Most users start with Code Review Mode — Debug Mode is optional for deeper investigation.

Who It’s For

Engineers who rely heavily on AI to write and fix code
Developers who want to move fast without breaking existing functionality

How to Use (Code Review Mode)

📹 Watch Latest Demo Video →

1. Enable Code Review Mode

Open Samurai Agent from the sword icon in the VS Code sidebar.
Go to Settings → Auto Regression Detection, then turn it on.

By default, Samurai Agent monitors all AI edits.
We recommend starting with this setting and adjusting sensitivity later if needed.

You can always dial it down — start strict, then relax.

Enable code review mode

2. Review AI Code Edits

Whenever AI tools like Cursor or Claude Code edit your files,
Samurai Agent detects potential regression risk and shows a warning in your IDE.

Click the warning to open Samurai Agent, then click once more to start the review.
The analysis runs automatically — just wait a moment.

Warning appears → click → start review

3. Fix the Regression

Once the review finishes, you’ll see a suggested fix at the bottom.

Use the Copy button to copy the generated response or prompt,
then paste it directly into Cursor or Claude to apply the fix.

Apply fix with AI

Advanced: Debug Mode (Optional)

1. Add relevant files and describe the bug

Add the files that are most relevant to the bug and briefly describe the issue.
Providing suspicious or related files significantly improves root cause accuracy.

Add files and describe bug

2. Run root cause analysis

Samurai Agent runs your code to identify the actual root cause using runtime information.
A visual summary helps you understand what’s happening at a glance.

3. Apply the fix with your AI coding tool

Copy the generated prompt and paste it into your AI coding agent (e.g. Cursor).
The prompt includes verified root-cause context, allowing the AI to double-check the cause before fixing the bug, which greatly reduces mistakes and regressions.

Tip: Using Cursor’s Plan mode works especially well, but any AI coding tool can be used.

Apply fix with AI

4. Review the fix to avoid regressions

After the AI applies the fix, return to Samurai Agent and click Review Fix.
Samurai Agent verifies that:

the root cause is actually resolved
no new regression bugs are introduced

Review fix

⚡ Quick Start (2 minutes)

📹 Watch Quick Start Tutorial →

How to Set Up

1. Install the extension

Search for “Samurai Agent” in the VS Code Marketplace (or any VS Code–compatible IDE) and install it.

2. Open Samurai Agent

Click the Samurai Agent icon in the VS Code sidebar, or
Press Cmd+Shift+P (Mac) / Ctrl+Shift+P (Windows/Linux), then type “Samurai Agent”

3. Try Samurai with no setup (free trial)

You can start using Samurai Agent immediately, without an API key or beta code.

Samurai Pro — deeper, runtime-based root cause analysis
(free trial · daily shared limit)
Samurai Lite — faster checks for simpler issues
(free tier · daily shared limit)

This lets you quickly see what Samurai’s root cause analysis looks like out of the box.

When the daily free-trial limit is reached, you can either try again later or unlock full access using one of the options below.

4. Unlock full access (recommended)

Option A: Add a beta code

If you have a beta code, add it in the Settings tab to unlock full access.

Add beta code

👉 Don’t have a beta code?
You can request one (and receive free credits) here:
https://samurai-agent.tech/

Option B: Add your own API key

You can also unlock Samurai by adding your own API key in the Settings tab.

Supported providers:

Gemini (recommended — Gemini Flash 3)
OpenAI
Anthropic

Supported Environments

Samurai Agent works with any JavaScript / TypeScript codebase, including but not limited to:

Node.js applications
Frontend frameworks: React, Vue, Angular
Full-stack frameworks: Next.js
Backend frameworks: Express
Monorepos and mixed JS/TS projects

If your project runs on JavaScript or TypeScript, Samurai Agent can analyze it.

🔒 Privacy & Data

Your code stays private

Samurai Agent acts as a root-cause context layer for AI coding agents.
Analysis is performed locally, and only the minimum required context is sent to your selected LLM provider.
Your source code and conversations are never stored or used for training by Samurai Agent.
Samurai Agent never modifies your files.

No fix or chat data stored

Samurai Agent does not store chat messages, fixes, or code diffs.
We collect only anonymous usage events (e.g. “analysis started”, “review fix clicked”) to improve product reliability.

You’re in control

Telemetry is optional and can be disabled at any time in the extension settings.

❓ FAQ

Does Samurai Agent modify or apply code fixes?

No.

Samurai Agent never edits your source code.
It provides root-cause context and reviews AI-generated fixes, but all code changes are applied by your AI coding tool (e.g. Cursor) or by you.

What is Samurai Agent best used for?

Samurai Agent is designed for:

Providing accurate root-cause context to AI coding agents
Reviewing AI-generated fixes to ensure the root cause is actually addressed
Catching regression risks before code is shipped
Situations where AI-generated fixes feel “correct” but break something else

When should I not use Samurai Agent?

Samurai Agent is not optimized for:

Simple syntax or type errors
Formatting or linting issues
One-line bugs with obvious stack traces

In those cases, direct AI fixes or linters are usually faster.

Does Samurai Agent replace AI coding agents like Cursor or Copilot?

No. Samurai Agent is complementary.

AI coding agents focus on writing code
Samurai Agent focuses on understanding root causes and reviewing fixes

Many users use Samurai Agent as a guardrail before merging AI-generated changes.

Does Samurai Agent execute my code?

Yes — only when needed to extract root-cause signals.

Execution happens in a controlled sandbox for analysis purposes. Your original files are never modified, and execution is limited strictly to context generation and fix review.

Is my code or fix data used for training?

No.

Samurai Agent does not use your code, fixes, or conversations for training and does not store them. Only minimal context is sent to your selected LLM provider, and data handling then follows that provider’s policies (e.g. OpenAI, Google, Anthropic).

💬 Support & Feedback

Samurai Agent is evolving quickly as a context and review layer for AI coding agents, and your feedback directly shapes its direction.

Questions, feedback, or feature requests
- Email: yuto@samurai-agent.tech
- X (Twitter): @yuto_samurai_ai

Samurai Agent — Prevent JS/TS Regressions from AI Edits

Samurai Agent