Quick Start Guide

Follow Installation first: create the Conda environment, install the UI/agent dependencies, configure both .env files, prepare storage, and obtain scientific resources for the tools you intend to use. Setup time depends on dependencies and model availability.

Start the web interface

From the OHMind source root, with its environment activated:

PYTHON="$(command -v python)" CHAINLIT="$(command -v chainlit)" bash start_OHMind.sh

Open http://localhost:8000 and sign in with your configured administrator credentials. Check backend liveness with curl --fail http://localhost:8005/health.

Start the terminal interface

PYTHON="$(command -v python)" bash start_OHMind_full.sh

This starts the MCP services and CLI. If services are already running, use PYTHON="$(command -v python)" bash start_OHMind_cli.sh instead.

Try a small request

Ask: “Calculate the molecular weight of aspirin from SMILES CC(=O)OC1=CC=CC=C1C(=O)O.” Inspect the actual tool result and any reported errors.

Once HEM model resources are available, ask for a supported backbone and cation type, then start a small optimization with five particles and three steps. Record the returned job ID and output directory; starting an asynchronous job is not confirmation that it finished.

Enable persistence separately

The default configuration disables durable state and recovery memory. For conversations that must survive backend restarts, follow Memory and persistence. Use the persistence tutorial to verify behavior before relying on it.

For interrupted HEM optimization, see Resume PSO. Database conversation checkpoints and optimizer checkpoints serve different purposes.