StarNet reviews keep asking one question, so here's my answer straight away: StarNet is a free, open-source AI agent harness that I rate 3.5 out of 5 after installing it and running real agents in it.
It's the smoothest and most fun office-style agent tool I've tested, and it's also not yet the tool I'd hand serious client work to.
StarNet puts your agents inside a pixel-art space station on your own computer, where you create them, give them permissions and watch them work.
This review is written from the operator's chair, so it follows the order you'd actually set it up and use it.
You'll get my verdict and scores first, then the install reality, then every part of the station I tested, then the honest limits.
If you want my team to look at how AI agents and content can get your business found in Google and AI search, you can book a free SEO strategy session.
My StarNet Verdict And Scores
My overall verdict is 3.5 out of 5.
Every score below is my own opinion from my own install, not a lab benchmark, so weigh it as one operator's view.
| What I scored | My score | Why it got that score |
|---|---|---|
| Fun and design | 10/10 | It's the most enjoyable agent interface I've used, and it makes you want to come back. |
| First-run setup | 8/10 | The first agent takes one or two clicks, but my source install needed an extra step. |
| Brain and model choice | 8/10 | You can sign in to existing subscriptions, paste a key or run a local model. |
| Permissions and control | 9/10 | Web and file access are clear switches on every agent. |
| Crew and multi-agent | 9/10 | Recruit, deploy and summon turn one agent into a visible team. |
| Chat speed | 6/10 | There's a slight delay every time you send a message. |
| Practical business use | 5/10 | I don't think there's a massive amount of practical use cases yet. |
| Maturity | 6/10 | The README itself calls it an early release. |
| Value for money | 9/10 | The code is free, but any paid model you connect still bills you. |
| Overall | 3.5/5 | It's fun, polished and free, with real rough edges for business work. |
That's the whole review in one table.
The rest of this page shows you where each score came from, so you can decide whether my reasons apply to you.
What StarNet Is, In One Minute
StarNet is a local-first desktop agent harness made by Andrew Sims, who publishes it on GitHub as androoAGI.
The repo's own description calls it "a living pixel-art station where real AI agents do real work".
It's open source under the MIT License, and the repo has over 1,500 stars on GitHub.
The README is clear that the station isn't a cartoon on top of nothing.
It says agents make real model calls, use real tools and create real costs.
It also says the interface must never show a state the harness can't prove, which is a rule I wish more agent dashboards followed.
I tested version v0.13.1, and the latest release on 16 October 2026 is v0.13.5.
If you've seen people call it "Startnet OS", that's the same tool, and I explain the name in my Startnet OS reviews post.
🔥 Want my exact StarNet Reviews setup, step by step? Inside the AI Profit Boardroom you get step-by-step video trainings, my installable Agent OS and four coaching calls a week with 3,400+ members building real automations. → Join the AI Profit Boardroom
Setup Reality: What Happened When I Installed It
This is the part most StarNet reviews skip, so I'll be specific.
There are two ways in.
The first route is the desktop app, which you download from the StarNet releases page for Windows or macOS.
The second route is running it from source with Node.js 18 or newer and Git.
I went the source route on my Mac.
The README says the sidecar runs on Node core modules only, so it should work without installing anything.
On my machine it still needed an npm ci before it would run, so budget a few extra minutes for that.
The README's default address is localhost on port 8787, and mine runs on port 8788.
Once it was running, the first screen asked me to create an agent, and from there it was fast.
| Setup step | What I did | What to watch for |
|---|---|---|
| Get the code | I cloned the repo from GitHub. | The desktop app skips this step entirely. |
| Install dependencies | I ran npm ci. |
My install needed it even though the README says zero-install. |
| Start the harness | I started the local web app. | Note the port it opens on. |
| Connect a brain | I signed in with Grok. | A paid model creates a real bill. |
| Create the first agent | I named it and picked a personality. | Choose the approval level carefully. |
The README is also honest about platforms.
It says Windows is the most-tested desktop target and macOS has less real-world coverage.
Linux packages are internal only and aren't a supported public release.
My full repo walkthrough is in my StarNet GitHub install guide if you want every command.
If you'd like my team to map out where AI agents fit your marketing before you spend time on tools, you can get a free SEO strategy session with us.
Creating My First Agent
You give the agent a name and look at its pixel-art appearance, which you can change.
You pick a personality, and I saw options like warm, dry, composed and unhinged.
You choose the approval level, which is either "ask for approval" or "full power".
Then you connect a brain by signing in or pasting an API key.
I signed in with Grok in one quick step.
You then pick the model and a reasoning level from low, medium, high or max.
I chose Grok 4.6 over Grok 4.7, because I actually prefer Grok 4.6.
The agent is created in one or two clicks, and a waking-up animation plays.
My agent said, "I'm awake. Let's give the station a direction."
That first-run moment is the best onboarding I've seen in an agent tool, which is why fun and setup score so well.
The Brains You Can Connect
StarNet doesn't ship its own model, so the brain you connect decides how smart your agents are.
In my video I explained that it can plug into existing subscriptions, including a Claude Code subscription through the Claude CLI, Kimi and Grok.
The README adds an OpenRouter key and sign-ins for supported Anthropic, OpenAI and Google accounts.
It says keys live in the OS keychain and never in the frontend.
You can also run it free with a local model through Ollama, and the README uses qwen3:8b as its example.
The README's honest caveat is that local models are slower and rougher on long tasks, and an 8B model takes about 10 GB of graphics memory.
That range of choice is why brains score 8 out of 10.
Running The Station Day To Day
The station view lets you zoom in and out, change the frame and go to a full zoom.
Everything you run lives in four docks, and you click them to open them, show everything or drag them around.
Your agents level up the more you use them, which makes it feel like a game.
I described it as a Tamagotchi for an AI agent, and I still think that's the best way to explain it.
The agent detail panel shows the model, the personality, the instructions and purpose, the access permissions and the approval prompts.
You can change the personality any time.
You can allow or remove web access and local file access for each agent.
The panel has Brief, Growth, History and Memories tabs, so you can see what an agent did and what it remembered.
That level of control is why permissions get 9 out of 10.
🔥 Want the agent setup I actually run my business on? Inside the AI Profit Boardroom, I've got step-by-step tutorials on Hermes Agent, agent dashboards and my installable Agent OS, where StarNet runs as one tab. You'll be learning alongside 3,400+ members with four coaching calls every week. → Get access here
Chat, Widgets, Apps And Skills
Chat lets you talk to any agent, and it shows the agent's online status and level.
My agent asked me, "What made you want to set up an AI agent?"
I told it SEO and gave it aiprofitboardroom.com.
The left-hand list switches between all your agents and shows which are idle and which are online.
You can manage projects and sessions from there as well.
There's a slight delay when you send a message, and that's the main reason chat scores 6 out of 10.
The plus button at the top adds widgets, including connected apps, pinned widgets, an app chooser and a daily brief.
You decide what information each widget shows.
The app catalog is labelled "add an ability", and it includes Gmail, Notion and GitHub.
Most apps sign in directly, and some need an API key.
There are real-world categories too, such as finding flights like a travel agent, sending physical letters, email and web search through Firecrawl.
The skill market has standard originals like ad copy testing and calendar scheduling.
Some skills are built in, some install later, and community updates can arrive over time.
The skill library holds frameworks, which are procedures the AI follows.
Examples include 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, Tasks And Quests
The Crew menu has crew agents and a Recruit option.
Recruit gives you roles like strategist, chief of staff, researcher and marketer, and you give each a name and an appearance.
Deploy overwrites an existing agent's setup with the new role.
Summon adds a new agent that stands next to yours in the station.
This is where StarNet feels most like a real team, which is why crew scores 9 out of 10.
Tasks, quests, a work list, building new stuff, the Systems menu and builds all live in one system.
It's ambitious, and it takes a while to learn how the pieces connect.
What Changed After My Test
I tested v0.13.1, and I haven't re-tested the newer builds, so my scores come from that version.
The v0.13.5 release notes, published on 16 October 2026, list several changes that target my complaints.
- The notes say you can now bring crew into a conversation just by asking your lead agent to add them.
- They say delegated work gets its own threads and one live card that shows who is doing what.
- They add a NEEDS YOU list that gathers every approval and question waiting on you in one place.
- They say building conveyor lines is now drag-and-drop, and finished jobs can be handed from one line to another.
- They say Claude Code runs are faster and cheaper.
- They say a skill an agent creates or changes now waits for a passing evaluation and your approval.
The notes also warn you to save a station backup before updating if you might ever go back to 0.13.3 or older.
I'm reporting these from the release notes, not from my own test, so check them yourself before you rely on them.
If you'd rather have my team look at how your site can win in Google and AI search while you test tools like this, grab a free one-to-one strategy session and we'll look at your site, your competitors and the keywords you're closest to winning.
StarNet Pros And Cons
| Pros | Cons |
|---|---|
| It's the smoothest office-style agent tool I've tested. | I don't see a massive number of practical use cases yet. |
| Agent creation takes one or two clicks. | There's a slight delay on chat messages. |
| Permissions are clear and per agent. | My source install needed npm ci. |
| The crew system makes a team visible. | It's an early release with less macOS coverage. |
| It's free and MIT licensed. | Paid brains still create real model bills. |
What Other People Asked About StarNet
The comments on my video were useful, so I'll pass them on.
Several viewers asked for more tutorials, which tells me people want to use it but aren't sure where to start.
One viewer was annoyed that they couldn't use it with Hermes Agent.
The README doesn't list Hermes Agent as a brain, but it does say StarNet can import an existing Hermes or OpenClaw agent from its home folder on disk.
It says the import brings across the persona, instructions, memory and model, while API keys must be re-entered.
It also felt like a much more fun version of Paperclip to me, and it reminded me of the first time I tested Hermes Agent.
StarNet Alternatives Worth Comparing
If you're reading StarNet reviews because you want an agent interface, these are the options I'd compare.
My own product is first on this list, so weigh that honestly.
| Rank | Name | Best for |
|---|---|---|
| 1 | Agent OS (my own, inside the AI Profit Boardroom) | Running Hermes, Claude and other agents from one dashboard with shared memory. |
| 2 | StarNet | A fun, visual crew you can watch and level up. |
| 3 | Herald OS | A free desktop-style interface with Hermes Agent at the centre. |
| 4 | Hermes3D | A 3D or 2D virtual office for agents on your own infrastructure. |
Agent OS is my dashboard for running a crew of AI agents under one shared memory, and members install it from inside the AI Profit Boardroom.
I actually run StarNet as a tab inside my Agent OS, rather than as my main tool.
Herald OS is a free, open-source project by Luke The Dev, and my Herald OS walkthrough shows how it works.
Hermes3D is another free MIT project by the same developer, and I cover it in my Hermes3D guide.
Who Should Run StarNet
StarNet is right for operators who want to learn how agent teams behave by watching one work.
It's right for creators and tinkerers who'll stick with a tool because it's fun.
It's not right yet for anyone who needs a hardened, client-facing automation system today.
If that's you, I'd run your real jobs elsewhere and keep StarNet as the fun side project.
If you want my team to build the plan for getting your business found in Google and AI search, book a free SEO strategy session and we'll map it out with you.
🔥 Want StarNet inside a full agent operating system? The AI Profit Boardroom has the Agent OS I built StarNet into, with tutorials, a member map and four weekly coaching calls where you can bring your setup. → Join the AI Profit Boardroom
Related Reading
- My StarNet AI agent setup walkthrough takes you through the first agent step by step.
- The StarNet GitHub install guide covers the repo, requirements and updates.
- My best UI for Hermes Agent ranking compares the interfaces I use around StarNet.
Also On Our Network
- 🌐 Is StarNet worth it for a business? My use-case review
- 🌐 StarNet reviewed with a full verdict, scores and pros and cons
- 🌐 Why I run StarNet as a tab in my Agent OS
- 🌐 The brains that make StarNet worth using, measured on GoldieBench
FAQ: StarNet Reviews
Is StarNet any good?
Yes, for fun and for learning how agent teams work, and I rate it 3.5 out of 5.
It loses points because it's an early release, chat has a slight delay and I don't see many practical business use cases yet.
Is StarNet free?
Yes, the code is free and open source under the MIT License.
Any paid model you connect still bills you, but you can run a local model through Ollama for free.
Is StarNet hard to set up?
The desktop app is the easiest route on Windows and macOS.
My source install needed an extra npm ci step even though the README says it runs without installing anything.
Can StarNet use Hermes Agent?
The README doesn't list Hermes Agent as a brain.
It does say StarNet can import an existing Hermes or OpenClaw agent, bringing across the persona, instructions, memory and model, and you re-enter API keys.
What is the latest StarNet version?
The latest release on 16 October 2026 is v0.13.5.
I tested v0.13.1, so my scores come from that version.
Is Startnet OS the same as StarNet?
Yes, "Startnet OS" is a common misspelling of StarNet.
My own video title calls it the StarNet Agent OS, which is probably where the "OS" part comes from.
If you'd like my team to look at how your content can rank in Google and AI search, book your free SEO strategy session here.
📺 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 SEO and link-building agency, Goldie Agency, with 70+ people in Bangkok.
- I've grown a YouTube channel to 400,000+ subscribers.
- I wrote the book "Link Building Mastery", which covers nine link building strategies.
- I run GoldieBench, where I score AI models on real one-shot builds.
→ Get my best AI training inside the AI Profit Boardroom
Fun, free and genuinely polished, with rough edges for business work, is my honest verdict across these StarNet reviews.
