Visitar URL original
Learn Command Code - Courses & Creator Videos
Command Code raises $5MRead more

//Learn

// Learn

Master Command Code, your way. Courses, creator videos, and guides. From your first prompt to shipping in production.

//

Videos

Ideas, shipped. See Command Code in the real world.

How did we make DeepSeek outperform Opus 4.7?

How did we make DeepSeek outperform Opus 4.7?

The story behind pushing DeepSeek past Opus 4.7 on real coding work with Command Code.

Ahmad Awais
Making DeepSeek v4 outperform Opus 4.7 with Taste

Making DeepSeek v4 outperform Opus 4.7 with Taste

How Command Code’s Taste system pushes DeepSeek v4 past frontier models on real coding work.

Latent Space
Fable 5 vs Opus 4.8 vs GLM 5.2: Can GLM 5.2 WIN?

Fable 5 vs Opus 4.8 vs GLM 5.2: Can GLM 5.2 WIN?

Head-to-head coding showdown pitting Fable 5, Opus 4.8, and GLM 5.2 on real tasks.

AICodeKing
China’s New AI Beat Claude AND OpenAI! (WHAT)

China’s New AI Beat Claude AND OpenAI! (WHAT)

Testing how China’s newest open models stack up against Claude and OpenAI on real coding work.

Mehul Mohan
I found the BEST AI For Coding (Open Source)

I found the BEST AI For Coding (Open Source)

A hands-on look at the best open-source AI coding agents and how they stack up.

Mehul Mohan
JUST $1 For Billions of Tokens?! (DeepSeek v4 Pro)

JUST $1 For Billions of Tokens?! (DeepSeek v4 Pro)

Putting cheap, high-volume token pricing to the test for real coding workflows.

Mehul Mohan
Explore More
//

Demos

Every feature, in under a minute. Press play.

Use /add-dir to work across multiple directories in Command Code.

Add more directories to a session so the agent can work across repos.

Command Code

Command Code supports Git worktrees.

Run parallel sessions side by side, each in its own worktree and branch.

Command Code

Stop copy-pasting errors. IDE diagnostics are quite strong.

Let the agent read your editor diagnostics and fix errors without copy-paste.

Command Code

Mods are more powerful than plugins.

Extend Command Code with mods built on the same harness it runs on.

Command Code

Fork your existing sessions and create new ones.

Branch a conversation to try another approach without losing the original.

Command Code

Connect to MCP servers to query data, use tools and automate workflows

Plug in MCP servers so the agent can reach your data, tools and services.

Command Code

Open an external editor for your multi-line and longer prompts.

Write long prompts in your own editor, then send them straight back.

Command Code

Plan review mode is best to comment, revise, and approve your plans.

Comment on, revise and approve a plan before any code gets written.

Command Code

Resume and pick up right where you left off. /resume command.

Reopen any past session with its full history and keep going.

Command Code

Paste images as your chat input in Command Code.

Paste an image into the prompt and the agent reads it as input.

Command Code

Make screenshots your prompts. No more describing just attach them.

Attach a screenshot instead of describing the UI or the bug in words.

Command Code

Try the new a11y mode with /design in Command Code

Review your interface for accessibility issues with the /design command.

Command Code

Writing tests have never been this easy.

Generate tests that match your codebase and run them in the same session.

Command Code

Pull requests are easier now.

Go from local changes to a written, ready-to-review pull request.

Command Code

Are code reviews blocking? No need to get slow!

Get a thorough review of your changes before a teammate ever looks.

Command Code

Think before you actually ship: Plan Mode!

Have the agent lay out its approach first so you can steer it early.

Command Code

Stop repeating the same prompt in every coding session. Do this instead!!

Save project rules once and every new session starts with them loaded.

Command Code

Does your code have your coding taste?

Command Code learns how you write code and keeps its output in your style.

Command Code

Save your favorite prompts forever and reuse them anytime!!

Turn prompts you use often into reusable commands you can call anytime.

Command Code

Do this to bring back any message or code instantly.

Rewind to an earlier message and restore the code from that point.

Command Code

Build your custom agents for your work

Define agents with their own instructions and tools for specific jobs.

Command Code

Does your AI has permission modes?

Choose how much the agent can do on its own, from ask-first to full auto.

Command Code

Your AI agents will now never forget their sessions and context.

Sessions and context persist, so agents pick up with everything they knew.

Command Code

Share your AI models usage with others. Tokenmaxing, cost, performance.

Share your model usage, tokens, cost and performance with others.

Command Code

Refactoring with Command Code hits different.

Restructure code across files while keeping behavior intact.

Command Code

Share your coding session with others.

Send a link to your session so others can see exactly what happened.

Command Code

Getting file context for your AI agent.

Point the agent at the right files with @ mentions for sharper answers.

Command Code

Understanding your codebase. (Next.js +TypeScript storefront)

Ask questions about an unfamiliar codebase and get a guided tour.

Command Code

Tool repairs and fixing tool calling issues in Command Code

Malformed tool calls get repaired automatically so sessions keep moving.

Command Code

/goal in Command Code

Set a goal and let the agent keep working until it is done.

Command Code
//

Guides

Short reads. Go deeper on taste, workflows, and shipping faster.

DESIGN

DeepSeek V4.1-Flash vs GLM 5.3 vs Kimi K3: we built the same bird flap game with all three

Three models built the same pipe-dodging bird flap game with Command Code /design. DeepSeek V4.1-Flash one-shot it clean at $0.0089, the lowest cost of the round - though not the lowest in this benchmark series overall.

Sep 11, 2026

DESIGN

Fable 5.1 vs GLM 5.3 vs Kimi K3: we built the same bird flap game with all three

One prompt, four models, zero hand-editing. A line-by-line read of what Fable 5.1, GLM 5.3, Kimi K3, and (updated) Qwen-3.8-Max actually shipped when asked to build the same pipe-dodging bird flap game with Command Code /design.

Sep 8, 2026

DESIGN

Claude Opus 5 vs GPT-5.6 Sol vs Grok 4.5: an endless-runner AI coding benchmark

Seven models built the same coin-collecting endless runner with Command Code /design. Opus 5 is the only one that renders real projected 3D - and the only one that forgot to save your high score.

Sep 6, 2026

DESIGN

We built the same neon brick breaker with 10 AI models - Opus 5 wins on craft, loses on one rule

Opus 5, DeepSeek V4 Pro, Fable 5, GLM 5.2, GPT-5.5, GPT-5.6 Sol, Kimi K3, Gemini 3.6 Flash, Laguna S 2.1, and Ling 3.0 Flash all built the same neon brick-breaker. Here is what a line-by-line code read found.

Sep 3, 2026

DESIGN

Muse Spark 1.1 vs Fable 5 vs Grok 4.5 vs GPT-5.6 Sol: a neon Snake game AI benchmark

Four models built the same neon Snake game with Command Code /design. We played all four side by side on desktop and mobile - Muse Spark 1.1 won, but not for the reason we expected going in.

Aug 31, 2026

DESIGN

Kimi K3 vs Qwen-3.8-Max: an elegant 3D photography portfolio, one floor cost 4-5x the other

Two models built the same minimalist photography portfolio landing page with a 3D floating-frame hero and lightbox gallery, using Command Code /design.

Aug 28, 2026
Explore More
//

Courses

Learn by building. Guided paths from your first command to production.

9 Lessons · 40 min

Introduction to Command Code

Learn to code with Command Code and build your coding taste.

4 Lessons · 15 min

Introduction to Taste

Teach your AI your coding taste so it ships code that feels like yours.