Instructions on running Dive a HOST application for LLMs with MCP support.
Use one of the install methods at Dive Ai Agent
Run dive like:
./Dive-0.7.2-linux-x86_64.AppImage --no-sandbox
conda create -n tools python=3.13.2
conda deactivate
conda activate tools
pip install uvx
Inspector is a great way to test it can run directly from npx:
npx -y @modelcontextprotocol/inspector npx server-postgres postgres://127.0.0.1/testdb
npx -y @modelcontextprotocol/inspector uvenv run mcp-server-git
Where npx -y @modelcontextprotocol/inspector
starts the inspector and
npx server-postgres postgres://127.0.0.1/testdb
executes the server
You should be able to run directly, like:
uvenv run mcp-server-git
or
node build/index.js
Open Dive > Tools Management > Edit Config Modify the JSON with any tools you like. Tip: Make sure the tools can run in the environment from where you started Dive.
Example of tools added in config.json. Make sure to modify YOUR_PATH, and if tools don't work modify $USER to your user name or set the env var.