OptiGen

AI-powered optimization modeling assistant built on LangGraph and Deep Agents. OptiGen guides users from problem formulation through schema design to executable solvers.

Quick Start (Usage First)

1. Install from PyPI:

   pip install optigen
   

2. Set environment variables (either export them or create a .env in the project root):

   # Required
   ANTHROPIC_API_KEY=your_anthropic_key  # Get your key at: https://console.anthropic.com/settings/keys
   
   # Optional (enables web search tools)
   TAVILY_API_KEY=your_tavily_key
   
   # Optional (tracing)
   LANGSMITH_TRACING=true
   LANGSMITH_API_KEY=your_langsmith_key
   LANGSMITH_PROJECT=optigen
   

3. Run the CLI:

   optigen
   

Features

• Problem Formulation (problem_formulator): Clarifies objectives and constraints before any schema or code is produced.
• Schema & Dataset Design (schema_dataset_designer): Creates request/response JSON schemas and example scenarios to match the agreed objectives/constraints.
• Solver Generation & Execution (solver_coder): Proposes solver strategies, uses available Python deps, and registers runnable entrypoints aligned to the schemas.
• Quick Start Mode: Can auto-build initial models for common problem types (e.g., VRP, scheduling, inventory) with transparent assumptions.
• Tool-Aware Workflow: Uses search (Tavily, optional), code execution, and dependency awareness to keep solutions consistent across steps.

Development (contributing)

Development Installation

For contributing or modifying the source code:

1. Clone the repo:

   git clone https://github.com/OptigenIO/OptiGen-core.git
   cd OptiGen-core
   

2. Install in editable mode:

   pip install -e .
   

Development Commands

• make test - Run unit tests
• make lint - Run linters and type checkers
• make format - Format code
• make help - Show all make commands