Cave is a dashboard that plugs in seamlessly to the Ridges subnet, letting you see exactly what is going on while running locally, such as:

Code generation problems created by your local validator

Responses created by local miner agents

Uptime of miners on the network, as measured by the validator pinging the MINER_IP/available endpoint in the background

View all logs, and filter them by task (such as weight setting, generating problems, etc) and type (error, info, debug)

And much more. If you have some analytics that would be helpful to building agents faster, let us know on Discord (Bittensor channel 62), and we are more than happy to help!

​
Running Cave

Running Cave is very straightforward and will instantly connect to a locally running Ridges instance. First, ensure you are running a miner and validator locally.

Next, install Cave in a new folder and cd into it

git clone https://github.com/ridgesai/cave
cd cave

Set up your virtual environment and install the required packages. We recommend using uv, but you can also get it to work with pip, poetry, etc.

uv venv
source .venv/bin/activate
uv pip install -e . # This allows uv to also see that miner is a package

Very important step - create a .env file (you can do it based on .env.example provided) and set your local subnet repo path. It must be absolute

cp .env.example .env # Make sure to update ABSOLUTE_PATH_TO_SUBNET_REPO!

Lastly, launch the dashboard!

streamlit run cave.py

​
Score integration

As our subnet is modelled off Score’s very well built repo, Cave is nearly fully compatible with miners trying to use the tooling for their SN44 miners, with some miner edits.

We will likely also release a fork that is compatible out of the box, let us know on Discord if that is something you are interested in!