StarNet GitHub Review: Every Feature Scored (Honest)

My StarNet GitHub verdict is simple: the repo at github.com/androoAGI/starnet is the most fun AI agent station I've tested, but it's a playground first and a productivity tool second.

StarNet is a free, open-source, local-first harness that puts your AI agents inside a pixel-art space station.

You create agents, give them a brain, and watch them walk around and do real work with real models and tools.

I installed it from the GitHub repo, ran it on my Mac, and went through every screen I could find.

This page is the reference version of that test.

You'll get the repo facts, the install steps, a scorecard for every feature, the problems I hit, and who I think should bother with it.

The Verdict On StarNet GitHub In One Table

Here's the short version before we go deep.

My overall rating is 3.5 out of 5.

It's a brilliant toy with a serious engine underneath, and the engine will matter more as the project matures.

What Is In The StarNet GitHub Repo?

StarNet is described in its repo as a local-first, gamified desktop harness for building and running real AI agent teams.

Here are the facts I checked directly in the repo on 7 October 2026.

The repo is split into a few clear parts.

The frontend folder holds the station world and desktop UI, the sidecar folder holds the local Node agent runtime, and the src-tauri folder holds the desktop shell.

There's also a large test folder and a QA folder with release receipts, which tells me the author takes testing seriously.

The README states one rule I really like, which is that the interface must never claim anything the harness can't prove.

How To Install StarNet From GitHub

Non-technical users should grab the desktop installer from the StarNet releases page.

There are files for Windows, Apple Silicon Macs and Intel Macs, and the install guide says Windows is the most-tested.

If you want to run from source, these are the commands that worked for me.

git clone https://github.com/androoAGI/starnet.git
cd starnet
npm ci
node sidecar/index.js

Then open http://localhost:8787 in your browser.

The README says the sidecar needs no install step, but my Mac needed npm ci before it would start properly.

I also run mine on port 8788 with PORT=8788 node sidecar/index.js, because another app already uses 8787 on my machine.

To update a source install, stop it, run git pull, run npm ci, and start it again.

🔥 Want to see how I wired StarNet into my own Agent OS? Inside the AI Profit Boardroom, I show how I run StarNet as a tab in my Agent OS, alongside my Hermes and Grok setups. You'll be learning with 3,400+ members and four coaching calls a week. → Get access here

StarNet Features Reviewed One By One

This section is a shorter tour than my full StarNet AI agent walkthrough, which goes screen by screen.

Agent creation: 9 out of 10

You name the agent, check its appearance and change it if you like.

You pick a personality from options that include warm, dry, composed and unhinged.

You choose ask for approval or full power, then connect a brain by signing in or pasting an API key.

It takes one or two clicks, and the waking-up animation is a lovely touch.

My agent woke up and said it was ready to give the station a direction.

Brain and model choice: 9 out of 10

I signed in with Grok in one quick step, which needs a SuperGrok or X Premium+ account.

You also pick the model and a reasoning level of low, medium, high or max.

I chose Grok 4.6 over Grok 4.7, because I prefer how 4.6 responds.

The repo also supports a Claude CLI sign-in through Claude Code, a Kimi sign-in, a ChatGPT sign-in, OpenRouter, Ollama and direct APIs from OpenAI, Anthropic, Gemini, xAI, Groq, Mistral, DeepSeek, Together, Fireworks, Perplexity and Cerebras.

That's the widest provider list I've seen in an agent office tool.

Station view: 8 out of 10

You can zoom in and out, change the frame and go to full zoom.

Everything you run lives in four docks, and you can click to open them, show everything and drag things around.

It looks great, and it's easy to get lost in it for longer than you planned.

Levelling up: 8 out of 10

Agents level up the more you use them, which turns using AI agents into a game.

It honestly feels like a Tamagotchi for an AI agent.

That sounds silly, but it does make you want to come back and use your agents.

Agent settings and permissions: 8 out of 10

The agent panel shows the model, the personality, the instructions and the access permissions.

You can allow or remove web access and local file access, and you can see approval prompts.

The Brief, Growth, History and Memories tabs tell you what the agent knows and what it has done.

You can change the personality at any time.

Chat and agent list: 6 out of 10

You talk to an agent in the chat, and it shows its online status and level.

The list on the left switches between all your agents, and you can manage projects and sessions.

I lost points here only because there was a slight delay each time I sent a message.

Widgets and app catalogue: 7 out of 10

The plus button adds widgets like connected apps, pinned widgets and a daily brief, and you choose what each one shows.

The app catalogue lets you add abilities such as Gmail, Notion and GitHub.

It also has real-world categories, including finding flights, sending physical letters, email and web search through Firecrawl.

Most apps sign in directly, but some need an API key.

Skill market and skill library: 7 out of 10

The skill market has original skills like ad copy testing and calendar scheduling, and some are built in while others install later.

The skill library holds procedures the AI follows, such as a decision framework for hard technical problems, making a plan, humanising content and handling a hard conversation.

You can enable or disable each framework.

Crew recruits: 8 out of 10

The crew screen lets you recruit roles like strategist, chief of staff, researcher and marketer, each with a name and appearance.

Deploy overwrites an existing agent with that role.

Summon adds a brand-new agent standing next to yours in the station.

Tasks, quests and systems: 6 out of 10

Tasks, quests, a work list, builds and a Systems menu all live in the same place.

It's powerful, but it took me a while to find my way around.

StarNet Vs Paperclip And Hermes

StarNet felt like a much more fun version of Paperclip to me.

Both tools organise agents into a team, but StarNet makes that team visible and playful.

It also reminded me of the first time I tested Hermes Agent, when everything felt new.

One viewer was annoyed that StarNet doesn't work with Hermes, and that's partly fair.

You can't run Hermes inside StarNet.

But the README says StarNet can import an existing Hermes or OpenClaw agent, including its persona, instructions, memory and model.

Your API keys don't come across, so you re-enter them in the KEYS tab.

The Problems I Hit With StarNet

The source install needed npm ci, even though the README says it doesn't.

Chat replies had a slight delay.

I don't see many practical use cases for most people yet, and I'd rather tell you that than oversell it.

The README is also honest that local models are slower and rougher on long tasks, and that an 8B model needs about 10 GB of graphics memory.

On the plus side, the v0.13.1 release fixed web requests that had been failing since 0.12.5, stopped crew freezing in the halls and made spending limits hold during retries.

If you want expert help deciding which AI tools deserve a place in your business, you can book a free strategy session with my team.

Who Should Install StarNet From GitHub?

You should install it if you love trying new agent tools and want a visual way to manage them.

You should install it if you already pay for Grok, Claude Code, Kimi or ChatGPT, because you can plug that sign-in straight in.

You should hold off if you need a battle-tested production system today.

You should also hold off if your computer can't run a decent local model and you don't want to pay for a hosted brain.

🔥 Ready to build agent workflows that actually ship work? The AI Profit Boardroom gives you my courses on Hermes Agent, Grok Bot, DeepSeek Harness and more, with four coaching calls a week to get unstuck. → Join the AI Profit Boardroom

Related Reading

• My DeepSeek Harness review scores another free agent harness so you can compare the two.
• The best AI models for coding helps you choose the brain you plug into StarNet.
• The full StarNet AI agent guide covers every station screen in more depth.

Also On Our Network

• 🌐 Installing StarNet from GitHub, step by step
• 🌐 Seven ways a business can use the StarNet repo
• 🌐 The StarNet quick start for Agent OS users
• 🌐 StarNet brain options ranked by real GoldieBench scores

FAQ: StarNet GitHub Review

Is the StarNet GitHub repo legit?

Yes, it's an MIT-licensed project by Andrew Sims at github.com/androoAGI/starnet.

It had over 1,100 stars when I checked, and the signed installers live on a separate releases repo.

Is StarNet worth installing?

It's worth installing if you want a fun, visual way to manage AI agents.

I don't think it has many practical uses yet, so don't treat it as your main workflow.

How does StarNet compare to Paperclip?

I'd call it a much more fun version of Paperclip.

Both organise agents into teams, but StarNet adds the station, levelling and recruits.

Can StarNet work with Hermes Agent?

You can't run Hermes inside StarNet.

You can import a Hermes or OpenClaw agent's persona, instructions, memory and model, but not its API keys.

What is the latest StarNet version?

The latest tag on 7 October 2026 was v0.13.1, released on 4 October 2026.

It was a repair release with fixes for web requests, group chats, crew movement and spending limits.

📺 Video notes + links to the tools 👉

🎥 Learn how I make these videos 👉

🆓 Get a FREE AI Course + Community + 1,000 AI Agents 👉

About Julian

I'm Julian Goldie, an SEO entrepreneur, author and founder of the AI Profit Boardroom, which has 3,400+ members.

I help business owners scale with AI agents, automation and SEO.

• I've built a 7-figure agency, Goldie Agency, with a team of around 50 people.
• I've grown a YouTube channel to 400,000+ subscribers.
• I wrote the Amazon best-sellers "SEO Link Building Mastery" and "Agency Marketing Mastery".
• My Udemy courses have taught over 50,000 students.

→ Get my best AI training inside the AI Profit Boardroom

It's fun, it's free and it's improving fast, and that's my honest verdict on StarNet GitHub.