Environment Variables

Complete reference for all environment variables used by OHMind, organized by category.

LLM Configuration
Workspace Configuration
Vector Database (Qdrant)
External Services
External Software Paths
Server Configuration
Application Settings
Example .env File
See Also

LLM Configuration

OHMind supports multiple LLM providers. Configure one of the following options:

OpenAI-Compatible APIs (Recommended)

Works with OpenRouter, Together AI, Groq, local models, and other OpenAI-compatible endpoints.

Variable	Description	Default	Required
`OPENAI_COMPATIBLE_API_KEY`	API key for the provider	-	Yes
`OPENAI_COMPATIBLE_BASE_URL`	Base URL for the API endpoint	-	Yes
`OPENAI_COMPATIBLE_MODEL`	Model name to use	`gpt-4o`	No
`OPENAI_COMPATIBLE_EMBEDDING_MODEL`	Embedding model name	-	For RAG

Example:

# Using OpenRouter
OPENAI_COMPATIBLE_API_KEY=sk-or-v1-your-key
OPENAI_COMPATIBLE_BASE_URL=https://openrouter.ai/api/v1
OPENAI_COMPATIBLE_MODEL=anthropic/claude-3.5-sonnet
OPENAI_COMPATIBLE_EMBEDDING_MODEL=openai/text-embedding-3-large

# Using Together AI
OPENAI_COMPATIBLE_API_KEY=your-together-key
OPENAI_COMPATIBLE_BASE_URL=https://api.together.xyz/v1
OPENAI_COMPATIBLE_MODEL=meta-llama/Llama-3-70b-chat-hf

# Using local Ollama
OPENAI_COMPATIBLE_API_KEY=ollama
OPENAI_COMPATIBLE_BASE_URL=http://localhost:11434/v1
OPENAI_COMPATIBLE_MODEL=llama3.1:70b

Direct OpenAI

Variable	Description	Default	Required
`OPENAI_API_KEY`	OpenAI API key	-	Yes
`DEPLOYMENT_NAME`	Model name	`gpt-4o`	No
`EMBEDDING_DEPLOYMENT_NAME`	Embedding model	`text-embedding-ada-002`	No

Example:

OPENAI_API_KEY=sk-your-openai-key
DEPLOYMENT_NAME=gpt-4o
EMBEDDING_DEPLOYMENT_NAME=text-embedding-3-large

Azure OpenAI

Variable	Description	Default	Required
`AZURE_OPENAI_API_KEY`	Azure OpenAI API key	-	Yes
`AZURE_OPENAI_ENDPOINT`	Azure endpoint URL	-	Yes
`OPENAI_API_VERSION`	API version	`2024-02-15-preview`	No
`DEPLOYMENT_NAME`	Deployment name	`gpt-4o`	No
`EMBEDDING_DEPLOYMENT_NAME`	Embedding deployment	`text-embedding-ada-002`	No

Example:

AZURE_OPENAI_API_KEY=your-azure-key
AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com/
OPENAI_API_VERSION=2024-02-15-preview
DEPLOYMENT_NAME=gpt-4o-deployment
EMBEDDING_DEPLOYMENT_NAME=embedding-deployment

Workspace Configuration

All computational results are organized under a unified workspace root.

Variable	Description	Default	Required
`OHMind_workspace`	Root workspace directory	-	Yes
`WORKSPACE_ROOT`	Alias for workspace root	`$OHMind_workspace`	No
`HEM_SAVE_PATH`	HEM optimization results	`$OHMind_workspace/HEM`	No
`QM_WORK_DIR`	ORCA QM calculations	`$OHMind_workspace/ORCA`	No
`MD_WORK_DIR`	GROMACS MD simulations	`$OHMind_workspace/GROMACS`	No
`MULTIWFN_WORK_DIR`	Multiwfn analysis outputs	`$OHMind_workspace/Multiwfn`	No
`CHEM_WORK_DIR`	Chemistry tool outputs	`$OHMind_workspace`	No

Example:

# Set the root workspace
OHMind_workspace=/data/ohmind_workspace

# Subdirectories (optional - defaults are derived from root)
HEM_SAVE_PATH=${OHMind_workspace}/HEM
QM_WORK_DIR=${OHMind_workspace}/ORCA
MD_WORK_DIR=${OHMind_workspace}/GROMACS
MULTIWFN_WORK_DIR=${OHMind_workspace}/Multiwfn
WORKSPACE_ROOT=${OHMind_workspace}

Workspace Directory Structure

When properly configured, the workspace will have this structure:

$OHMind_workspace/
├── HEM/                    # PSO optimization results
│   ├── best_solutions_*.csv
│   ├── best_fitness_history_*.csv
│   └── optimization_*.log
├── ORCA/                   # QM calculation files
│   ├── temp_*/             # Per-job temporary directories
│   └── results/            # Preserved results
├── GROMACS/                # MD simulation files
│   ├── *.pdb, *.top        # Topology files
│   └── *.xtc, *.trr        # Trajectories
├── Multiwfn/               # Wavefunction analysis
│   └── <job-name>/         # Per-analysis directories
└── qdrant_db/              # Vector database (if local)

Vector Database (Qdrant)

Configuration for the RAG system’s vector database.

Variable	Description	Default	Required
`QDRANT_URL`	Qdrant server URL	-	For server mode
`QDRANT_PATH`	Local storage path	`$OHMind_workspace/qdrant_db`	For local mode
`QDRANT_API_KEY`	API key (if required)	-	No

Local Mode (Recommended for single-user):

# Use local file-based storage
QDRANT_PATH=/data/ohmind_workspace/qdrant_db
# Leave QDRANT_URL unset

Server Mode (For multi-user or production):

# Connect to Qdrant server
QDRANT_URL=http://localhost:6333
QDRANT_API_KEY=your-qdrant-api-key
# Leave QDRANT_PATH unset

External Services

Web Search (Tavily)

Variable	Description	Default	Required
`TAVILY_API_KEY`	Tavily API key for web search	-	For web search

Get a free API key at tavily.com.

TAVILY_API_KEY=tvly-dev-your-key

HuggingFace

Variable	Description	Default	Required
`HUGGINGFACE_TOKEN`	HuggingFace access token	-	For reranking

HUGGINGFACE_TOKEN=hf_your_token

LangSmith (Tracing)

Variable	Description	Default	Required
`LANGCHAIN_TRACING_V2`	Enable LangSmith tracing	`false`	No
`LANGCHAIN_API_KEY`	LangSmith API key	-	If tracing enabled
`LANGCHAIN_PROJECT`	Project name for traces	`hem-design-agents`	No

LANGCHAIN_TRACING_V2=true
LANGCHAIN_API_KEY=ls_your_key
LANGCHAIN_PROJECT=my-ohmind-project

External Software Paths

Paths to external computational chemistry software.

Variable	Description	Default	Required
`OHMind_ORCA`	Full path to ORCA binary	-	For QM calculations
`OHMind_MPI`	Directory containing MPI binaries	-	For parallel ORCA
`MULTIWFN_PATH`	Full path to Multiwfn executable	-	For wavefunction analysis

Example:

# ORCA quantum chemistry
OHMind_ORCA=/opt/orca/orca
OHMind_MPI=/usr/local/openmpi/bin

# Multiwfn wavefunction analysis
MULTIWFN_PATH=/opt/multiwfn/Multiwfn

# GROMACS (typically on PATH after sourcing GMXRC)
# source /usr/local/gromacs/bin/GMXRC

Verifying External Software

# Check ORCA
$OHMind_ORCA --version

# Check Multiwfn
$MULTIWFN_PATH <<< "q"

# Check GROMACS
gmx --version

Server Configuration

Settings for the FastAPI backend server.

Variable	Description	Default	Required
`API_HOST`	Server bind address	`0.0.0.0`	No
`API_PORT`	Server port	`8000`	No
`ALLOWED_ORIGINS`	CORS allowed origins (comma-separated)	`http://localhost:3000,http://localhost:3001`	No

Example:

API_HOST=0.0.0.0
API_PORT=8005
ALLOWED_ORIGINS=http://localhost:3000,http://localhost:8000,https://myapp.example.com

MCP Configuration

Variable	Description	Default	Required
`MCP_CONFIG_PATH`	Path to mcp.json configuration	-	Yes

Example:

MCP_CONFIG_PATH=/path/to/OHMind/mcp.json

Application Settings

General application behavior settings.

Variable	Description	Default	Required
`LOG_LEVEL`	Logging verbosity	`INFO`	No
`MAX_TOKENS`	Maximum tokens for LLM responses	`4096`	No
`TEMPERATURE`	LLM temperature (0.0-1.0)	`0.7`	No

Example:

LOG_LEVEL=DEBUG
MAX_TOKENS=8192
TEMPERATURE=0.5

Example .env File

Complete example configuration:

# =============================================================================
# OHMind Environment Configuration
# =============================================================================

# -----------------------------------------------------------------------------
# LLM Configuration (choose one provider)
# -----------------------------------------------------------------------------

# Option 1: OpenAI-Compatible API (recommended)
OPENAI_COMPATIBLE_API_KEY=sk-your-api-key
OPENAI_COMPATIBLE_BASE_URL=https://api.provider.com/v1
OPENAI_COMPATIBLE_MODEL=gpt-4o
OPENAI_COMPATIBLE_EMBEDDING_MODEL=text-embedding-3-large

# Option 2: Direct OpenAI
# OPENAI_API_KEY=sk-your-openai-key

# Option 3: Azure OpenAI
# AZURE_OPENAI_API_KEY=your-azure-key
# AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com/

# -----------------------------------------------------------------------------
# Workspace Configuration
# -----------------------------------------------------------------------------

OHMind_workspace=/data/ohmind_workspace
WORKSPACE_ROOT=${OHMind_workspace}
HEM_SAVE_PATH=${OHMind_workspace}/HEM
QM_WORK_DIR=${OHMind_workspace}/ORCA
MD_WORK_DIR=${OHMind_workspace}/GROMACS
MULTIWFN_WORK_DIR=${OHMind_workspace}/Multiwfn

# -----------------------------------------------------------------------------
# Vector Database (Qdrant)
# -----------------------------------------------------------------------------

# Local mode (recommended for single-user)
QDRANT_PATH=${OHMind_workspace}/qdrant_db

# Server mode (uncomment for multi-user)
# QDRANT_URL=http://localhost:6333
# QDRANT_API_KEY=

# -----------------------------------------------------------------------------
# External Services
# -----------------------------------------------------------------------------

TAVILY_API_KEY=tvly-dev-your-key
HUGGINGFACE_TOKEN=

# -----------------------------------------------------------------------------
# External Software Paths
# -----------------------------------------------------------------------------

OHMind_ORCA=/opt/orca/orca
OHMind_MPI=/usr/local/openmpi/bin
MULTIWFN_PATH=/opt/multiwfn/Multiwfn

# -----------------------------------------------------------------------------
# MCP Configuration
# -----------------------------------------------------------------------------

MCP_CONFIG_PATH=/path/to/OHMind/mcp.json

# -----------------------------------------------------------------------------
# Server Configuration
# -----------------------------------------------------------------------------

API_HOST=0.0.0.0
API_PORT=8000
ALLOWED_ORIGINS=http://localhost:3000,http://localhost:3001

# -----------------------------------------------------------------------------
# Application Settings
# -----------------------------------------------------------------------------

LOG_LEVEL=INFO
MAX_TOKENS=4096
TEMPERATURE=0.7

# -----------------------------------------------------------------------------
# LangSmith Tracing (optional)
# -----------------------------------------------------------------------------

LANGCHAIN_TRACING_V2=false
LANGCHAIN_API_KEY=
LANGCHAIN_PROJECT=hem-design-agents

Environment Variable Precedence

When the same variable is defined in multiple places:

Shell environment (highest priority)
Component-specific .env (OHMind_agent/.env, OHMind_ui/.env)
Root .env (lowest priority)

# Example: Override for a single command
OHMind_workspace=/tmp/test_workspace python -m OHMind_cli

Environment Variables

Table of Contents

LLM Configuration

OpenAI-Compatible APIs (Recommended)

Direct OpenAI

Azure OpenAI

Workspace Configuration

Workspace Directory Structure

Vector Database (Qdrant)

External Services

Web Search (Tavily)

HuggingFace

LangSmith (Tracing)

External Software Paths

Verifying External Software

Server Configuration

MCP Configuration

Application Settings

Example .env File

Environment Variable Precedence

See Also