StartupHakk/OpenMonoAgent.ai

Second the huge utility of being able to highlight-copy text directly generated by the agent.

@rcraig12 - have you tried using it to reference files at a path? Remember that these are working in a docker / sandbox - so you have to reference inside the working directory. We can take a look as well.

We are working on improving it. Copy and Paste in the CLI is going to be tricky. We are working on a VC Code plugin - that should help a ton.

@pablo7913 @xerudro - pull down the latest. We have added dedicated support and it works great. The accuracy is a bit lower than a 24GB VRAM GPU, the CPUs we support that have 32GB VRAM. Pull down the latest.

Hello,

Previously, the system required 24GB VRAM minimum, which caused llama-server issues on 12GB and 16GB GPUs (like restart loops during model loading and health check failing).

**Exciting update!** We have now improved the install script with better hardware detection and handling. It no longer assumes a fixed setup and will now prompt you to choose the appropriate GPU mode/model tier when VRAM is detected, making installs more reliable across different systems.

We’ve also added dedicated support for 12GB–16GB GPUs (RTX 3060, 4060 Ti, 4080, etc.) with optimized models and loading strategies. While these won’t match 24GB+ GPU performance (3090/4090/5090 class), they’ve been tested and are now stable and usable with correct configuration.

Would love for you to try the latest version and share feedback!

having the same issue on macbook pro m5 32gb ram (not sure if it can even run on this machine) - but a model does not load when running openmono agent after setup and I see MODEL_NAME variable is not set when running openmono start.

> Use my fork to run the agent natively on Windows. The LLM server still needs to be running using WSL or using any other server setup (LM Studio, etc.). Works for me. https://github.com/AndyHen/OpenMonoAgent.ai

So basically, build it on Windows and run?

I wish there would be a release for this via winget or something...

Windows 11, using WSL. Issue present in Windows Terminal PowerShell, then typing WSL to open.
Issue also appears when selecting 'Ubuntu' directly from Windows Terminal.
Issue also appears when using MobaXTerm and selecting 'Ubuntu WSL'.

Observation:
Cannot use mouse to select contents of CLI.
Example:
Agent trying to help me resolve configuration issue with a website, and was providing blocks of strings to change for NGINX config. Multiple iterations while troubleshooting. Approx 5-12 lines with keys each time.

Expectation:
I can highlight text generated by the agent, and copy it into other Windows tools from the clipboard.

Actual Result:
No text can be highlighted. Must export entire history, open Notepad, find relevant text, and copy it out.

Why this is a deal breaker:
Work tasks regularly require copying data generated out of the CLI. Having to type short strings is tolerable, but the lack of this feature will relegate the solution to a small niche of my work, rather than ...

Thanks for your reply. When i enter the first sudo command, after entering my password, i get `"sudo: nvidia-ctk: command not found"`

Try running

```
sudo nvidia-ctk runtime configure --runtime=docker
sudo systemctl restart dock
```

i've tried the config flags and command line options in my troubleshooting. The settings.json file gets updated with the model I specify but then I still get the same error when trying to run the agent. I'm sure it's me. Please help!

This is content from the log when I tried to start the agent:

[2026-05-07 21:04:13.747] [INFO] Session starting — model= endpoint=http://llama-server:7474 workdir=/workspace
[2026-05-07 21:12:42.261] [DEBUG] LLM request: model= messages=2 tools=19 endpoint=http://llama-server:7474
[2026-05-07 21:12:43.311] [WARN] LLM retry 1/3 after 1s
[2026-05-07 21:12:47.322] [WARN] LLM retry 2/3 after 4s
[2026-05-07 21:13:10.260] [WARN] LLM retry 3/3 after 16s
[2026-05-07 21:14:51.205] [ERROR] LLM connection failed
[2026-05-07 21:14:51.269] [ERROR] System.Net.Http.HttpRequestException: Resource temporarily unavailable (llama-server:7474)
---> System.Net.Sockets.SocketException (11): Resource temporarily unavailable
at System.Net.Sockets.Socket.AwaitableSocketAsyncEventArgs.ThrowException(SocketError error, CancellationToken cancellationToken)
at System.Net.Sockets.Socket.AwaitableSocketAsyncEventArgs.System.Threading.Tasks.Sources.IValueTaskSource.GetResult(Int16 token)
at System.Net.Http...

Hi, i have the same problem with a RTX 4070 SUPER 12GB. Llama-server never starts.

Use my fork to run the agent natively on Windows. The LLM server still needs to be running using WSL or using any other server setup (LM Studio, etc.). Works for me.
https://github.com/AndyHen/OpenMonoAgent.ai

please need install tutorial using ai but nothing really working